The official DANCAD3D (tm) Beta Test Web site.

FEATURE FILM 4K+ DIGITAL CINEMA UNCOMPRESSED EDITING AND SOUND MIXING
PLUS DIY MOVIE 35mm FILM SCANNING AND DIY 35mm FILM RECORDER SOFTWARE.
"FREEISH" DI (DIGITAL INTERMEDIATE) SOFTWARE TO DOWNLOAD FOR MAKING
FEATURE MOTION PICTURES FOR CINEMA THEATRE, ULTRA-HD, HD, AND BROADCAST.
WORKS WITH FOOTAGE FROM DIGITAL CINEMA CAMERAS AND MOVIE FILM SCANS.
HIGH DEPTH COLOR CORRECTION. FRAME ACCURATE EDITING. HI-FI SOUND MIXING.

Click on these thumbnails for related information about my current software, features, and IM support. Hold [Shift] and click [Reload] to refresh IM thumbnail.

If are helping "Beta Test" you should report bugs and feel free to ask questions about the program's commands, it is best to e-mail me at tempnulbox (at) yahoo (dot) com and put "DANCAD3D (tm) 24x7 SUPPORT SUBMISSION" in the email subject line so your mail is not deleted as junk. See Section: 8 for more information about support related issues. I want you to ask questions so long as they are on the subject and relate to the current program's commands. If you do not get some kind of reply assume I did not get your message and resend. All submissions and correspondence become the sole property of Daniel H. Hudgins to do with as he sees fit, so stay on subject.

Digital Cinema de-Bayer and Image Processing info from DANCINEC.ZIP (tm).
This Section, About DANCINEC.ZIP (tm), and Text from DANCINEC.TXT.

DOCUMENT: DANCINEC.TXT Copyright (C) 2011 by Daniel H. Hudgins, All Rights Reserved. Terms of use: This "Beta Test" document may only be used in accord and within the limitations imposed by the current End User License Agreement "EULA" posted at the author's Web site www.DANCAD3D.com (sm) in file S0000000.HTM, any other use or copying is prohibited. This document is provided "AS IS" without warranty of any kind express, expressed, or implied. Mistakes, errors, and omissions should be reported according to the instructions in SECTION: 8 of the current "On-Line" version of my Web site www.DANCAD3D.com (sm). Any trade marks mentioned in this document, this web site, or the programs menus and such, belong to their respective owners, and are mentioned not as an endorsement or recommendation but rather for informational and educational purposes to enrich the discussion of the matters dealt with. This preliminary document has some brief descriptions of changes made to my Cinema de-Bayer and image processing programs DANCINED.EXE (tm), DANCINEC.EXE (tm), DANCINEW.EXE (tm). This file is meant to be included in the initial "Beta Test" v0.01 distribution to users acquaint themselves with some of the many commands and features in the programs. Other text may be located at my "Beta Test" Web site www.DANCAD3D.com (sm). Please read this document before beginning to use the programs. This document is not a complete list of changes made to the programs, and may not reflect the operation of the version of the program accompanying it in all respects. The programs are in preliminary development and testing and may be still undergoing change, so the results obtained from any of the commands may be different than expected, and the operation of older commands may have changed as well. Since so many changes have been made to the programs you should not expect any of the commands to operate as you have used them in the past, and you should frequently back-up and save what you are working on so that you do not lose everything when the program crashes. All specifications, descriptions, and instructions are subject to change without notice. Be sure to see also the file INFO16.TXT that is in archive DANCAD16.ZIP (tm). DANCAD87.EXE (tm) is in archive DANCAD16.ZIP (tm) also, you will need it in addition to DANCINEC.EXE (tm) in DANCINEC.ZIP (tm) for some operations. See also DANCINEL.EXE (tm) for making a 35mm film recorder to transfer frames processed in DANCINEC.EXE (tm) to movie film for projection in movie theaters, and DANCINES.EXE (tm) for DIY 8mm, S8, 16mm, Super16, Techniscope (tm), 35mm, Super 35mm, VistaVision (tm), and 65/70mm or IMAX (tm) movie film scanning to make frames for processing in DANCINEC.EXE (tm). I would like to thank the thousands of users of my programs who have helped "Beta Test" the many revisions of my programs since about 1986, I hope you will enjoy checking out some of the newer program features that I have spent years working on. Best wishes for success in your projects. If you have problems getting the programs to work for your filmmaking project or find what seem to be bugs, please email me with the name of the program of mine that your using in the subject line, by email address should be at www.DANCAD3D.com (sm) at the top of each page and in Section: 8. Be sure to put "DANCINEC.EXE" in the subject line of the email so it does not get deleted as junk mail. If you don't get a reply about filmmaking software questions it probably means that I did not get your email or your spam filter blocked or deleted my reply, so try again using another email account. --- DNG NOTICE Although how "Compliant" my DNG code might be in Adobe (tm)'s eyes, is unknown at the time I am writing this, I am including their notice here. I do not support all DNG or TIF Tags and disregard many in order to apply the color model used in my de-Bayer code that can make non-linear adjustments to the transfer curves to avoid things like pink highlights due to non-linear response of the sensor, sensor pre-amps, ADC near maximum signal and such, as well as the general differences between how I am processing the DNG image data and how programs that use the Adobe (tm) SDK might process it. However, any Bayer type DNG file (Not so called Linear DNG which are a form of non-white-balanced TIF RGB image file (I don't know if you can rename such Linear DNG as 48bpp TIF and load them that way, it would depend on what tags are used in them and if they are 16bpc etc.)) should be able to be processed to produce a "usable" image with the right adjustments to the KCC files and other de-Bayer settings, if you find DNG files that to not load and process well you can try to contact me for support so that I can look into what tag or tags are causing an issue in your DNG file (I would need sample files probably), as some tags and how they are implemented may vary quite a bit depending on what program or camera made the DNG file that you are trying to load and process in my programs. The programs I have written that read and write DNG files are: DANCINED.EXE (tm), DANCINEC.EXE (tm), and DANCINEW.EXE (tm). Those image processing programs come with DANGUIVU.EXE (tm) for its graphics display, at the time I am writing this DNG files are converted to binary image data before being transferred to DANGUIVU.EXE (tm), although some future version might read DNG files directly. Those programs should always come with this document in the same DANCINEC.ZIP file. The following is a quote from the Adobe (tm) web site: Notice Any Compliant Implementation distributed under this license must include the following notice displayed in a prominent manner within its source code and documentation. "This product includes DNG technology under license by Adobe Systems Incorporated." More information may be found at: http://www.adobe.com/products/dng/ end quote. I do not use the Adobe (tm) SDK code in my programs. The DNG reading and writing programs associated with this document DOS and Windows (tm) versions were written in TMT Pascal (tm) and use its units. The DOS version has a notice for its 32bit loader in its code. --- FORWARD I have written DANCINEC.EXE (tm) et al. to complement my DI/NLE/MIX/SCAN/FILMOUT system that is made up of DANCAD87.EXE (tm), DANCINES.EXE (tm), and DANCINEL.EXE (tm). DANCINEC.EXE (tm) adds or improves these capabilities: 1) De-Bayer of "True RAW" Digital Cinema Camera data for Cinema quality film replacement un-compressed shooting. Both fast "Workprint" and higher quality "Finish" processing methods are supported. The de-Bayer code supports automated "bad pixel removal" and "fixed pattern noise" or FPN subtraction. 2) Faster production of the PIX frames used for editing sync playback. 3) DNG conversion camera format to BMP/CIN/DPX/TIF are supported in the de- Bayer processing methods. 4) CIN and DPX input and output is supported with "Log" to monitor gamma and the reverse supported. DPX are supported for both 10bpc and 16bpc. Conversion from DPX to TIF and TIF to DPX allow interfacing other cameras or workflows with DANCAD87.EXE (tm) for un-compressed DI using the film industry's standard format DPX. 5) RGB to RGB process methods for de-Noise and color correction and Grading on frames made with compressed cameras (like RED (tm) or Canon (tm)) that have their own de-Bayer process. 6) "OPTICALS" processed un-compressed up to 48bpp for titles, compositing, fades, etc. 7) SMPTE time code support from DPX and some RAW meta-data. 8) Burn-in and Window-dub of guide elements, SMPTE time code, Time and Date information from frames, titles, and sub-titles. This includes commands to convert Vector font files used in DANCAD87.EXE (tm) into Raster font files used by DANCINEC.EXE (tm) to burn in the text and numbers etc. in various languages. 9) Other tasks such as file header trace for de-Bug etc. 10) "Sync" features to the Edit list made with DANCAD87.EXE (tm) to help automate some functions and allow for color correction of just the frames edited into the finished film (for a kind of final color correction conform pass based on the editing done with the "proxy" PIX frames). 11) Experimental features such as "reverse de-bayer" to RAW or DNG. 12) Delta-encoding of True RAW or Bayer DNG data to reduce disk space needed for archive, something like using ZIP but made just for camera sensor data so that it is fully lossless yet still can reduce disk space maybe 20 to 25% depending on the images in the frames and the exposure and noise levels. 13) Greater speed processing various frames used in the movie's production over what DANCAD87.EXE (tm) could do, because DANCINEC.EXE (tm) is 32bit code that can use more of the computers memory for frame buffers so less disk I/O is needed during some processing steps. This can be a major gain given the processing times required for an un-compressed 4K DI. 14) DANCINEC.EXE (tm) and its other versions DANCINED.EXE (tm) and DANCINEW.EXE (tm) help overcome some OS issues that Microsoft (tm) has introduced, so you can hopefully batch process frames on more current OS versions and take advantage of faster and larger HDD or SSD arrays to speed frame processing. 15) A probe in the graphics preview helps with checking the D-LAD (tm) reference levels for conversions from monitor gamma viewable files like BMP/TIF to so called "Log-C" or "Cineon (tm)" CIN/DPX files to reduce workflow issues with the reference levels not being at the right code values in the files made or read and converted. The input and output LUT would be adjusted to get the probe to read the right reference levels off the graphics display of the image frames. These additions improve the usability of the DI system, and expand its capabilities for lower cost motion picture production by a considerable degree. The motion picture industry is converting from film based production and projection to an all digital workflow from camera to projector. My DI system can be used with both film based production and all Digital based productions. My de-Bayer is designed for batch processing of Digital Cinema Camera frames intended for use in motion picture productions. DPX output, and input, are now supported for easier interfacing to additional third party motion picture programs for VFX, DCP, and other "external" work on frames as may be needed. I hope you find DANCINEC.EXE (tm) et al. a useful addition to my other DI system programs. If you encounter any issues please report them by email with the name of the program involved in the email subject line. Read Section: 8 at http://www.DANCAD3D.com for guidelines about support, and to find my current email address. If you don't get a reply please try again. --- OVERVIEW OF PROGRAMS IN DANCINEC.ZIP (tm) To keep the text from getting too long, if you see the program name DANCINEC.EXE (tm) it is also talking about DANCINED.EXE (tm) and DANCINEW.EXE (tm) most of the time. If you see DANCAD87.EXE (tm) it is also talking about DANCAD3D.EXE (tm) most of the time. There are more than one version of the more or less same program to make up for issues with the OS or hardware, but the menus and program functions are similar. Please use the most up-to-date revisions of the programs in your "Beta Testing" so that you do not report or experience bugs that have already been dealt with. Three versions of the same de-Bayer and image processing program are included in DANCINEC.ZIP (tm) the hope that one of them might be usable under the version of OS you are using (mostly because of Windows (tm) induced issues). These programs are for three tasks primarily: 1) To de-Bayer RAW Digital Cinema Camera frames into color corrected RGB image files for editing. 2) To color correct RGB to RGB Cinema image frame files made by compressed Digital Cinema Cameras that have their own de-Bayer software but can make BMP/CIN/DPX/TIF image frames. 3) To do "opticals" using RGB frames that have been de-Bayered or color corrected for editing into Cinema projects. The program is designed to be used for batch processing of image frames on several computers at once. The programs are for working with Cinema size image frames such as 4K, 2K, 1920x1080p and other sizes, and process images that are up to 48bpp and un-compressed for use with my DI/NLE/MIX system in DANCAD87.EXE (tm). This DANCINEC.TXT file should have been found in DANCINEC.ZIP file. This TXT documentation is about three similar programs named: DANCINED.EXE (tm) The 32bit DOS version, runs under Windows ME (tm) full install and XP home (tm) (but not the 16bit version of FreeDOS (tm) it seems, there is a development of a 32bit version of FreeDOS (tm) but that has not been tested as of the time I am writing this). The 32bit DOS version has options for full screen graphics using VESA VBE 1.2, 2.0, to 3.0 graphics, as well as using the DANGUIVU.EXE (tm) image viewer as well as just about any other graphics program that can read a 24bpp BMP sample image. DANCINEC.EXE (tm) The Windows Console (tm) version for ME (tm), XP home (tm), and maybe later Windows (tm) versions. The Windows Console (tm) version should be tried first as it may work best and has options for Full Screen graphics using DIRECT-X (tm) or using the DANGUIVU.EXE (tm) image viewer as well as just about any other graphics program that can read a 24bpp BMP sample image. DANCINEW.EXE (tm) The Windows GUI (tm) version for XP home (tm) and maybe Windows 7 (tm). The Windows GUI (tm) version uses the Windows (tm) file folder directory pop-up. The Windows GUI (tm) version does not support full screen graphics, but the values passed to DANGUIVU.EXE can make the graphics window equal to the desktop screen width, and the image height can be the desktop height minus 13 pixels, so its almost full screen, or smaller as you set it up in the graphics part config file setup. You can also output the graphics preview for checking the results to show up in any program that can take a 24bpp BMP filename as a command line parameter (drag-and-drop) so that the graphics viewer program will open automatically and then DANCINEW.EXE will continue automatically when you close the viewer program. If you see any notes saying that the GUI version does not have a graphics display, those are not longer up-to-date as the external GUI viewers are now supported automatically. The DOS version supports a graphics preview to check color corrections and other image issues that uses the video boards VBE 1.2, 2.0, or 3.0 VESA BIOS for graphics modes, and should work up to 2048x1536x32 on video boards that can display that high. There may be problems on some computers getting that VESA graphics to work under windows XP home (tm) at resolutions over 640x480, that may be do to having set the composite video out on the video board to enable, if you disable the composite video out that might resolve that issue. Under Windows ME (tm) things should work up to what the video board (graphics card) BIOS supports under VBE. One issue with Windows (tm) versions after ME (tm) is that they may set the video refresh rate too low for the monitor to work at, so the monitor will say Hz out of range error (it seems Windows XP Home (tm) (and later?) prevents the user program from re-setting the monitor Hz to a valid value through VESA). For computers using video monitors that are not the same shape as the VESA modes supported by your graphics card, you should select using DANGUIVU.EXE (tm) as the image viewer, at least if you are using Windows (tm) as the OS, or you would need a monitor that works with your graphics card in VESA modes. The default graphics are VESA for the 32bit DOS version. The Windows Console (tm) version of the program looks and works like the DOS version, but should display the video modes up to 1600x1200 even when the DOS version is limited to 640x480, if your video board supports 1600x1200 under Direct-X (tm). Direct-X (tm) needs to be loaded on your computer and the Windows desktop loaded in order for the windows console version to work. The Windows CONSOLE version is the preferred version to use on OS later than ME (tm). For computers using video monitors that are not the same shape as the DIRECT-X (tm) modes supported by your graphics card and the graphics unit the program is compiled with, you should select using DANGUIVU.EXE (tm) as the image viewer. DANGUIVU.EXE (tm) opens a GUI window on the desktop from the Windows Console (tm) version whose window goes blank until you quit DANGUIVU.EXE (tm). The default graphics are DANGUIVU.EXE (tm) for the Windows Console (tm) version. The Windows GUI (tm) version is only supplied for use in case the Windows Console (tm) version will not work on the Windows (tm) version you are using. The Windows GUI (tm) version does not support VESA or DIRECT-X (tm) graphics itself, but uses the external DANGUIVU.EXE (tm) program that you can set up to the desktop width and to 13 pixels less than the desktop height. The Windows GUI version uses the WinCRT unit in the compiler I am using currently, and may show some anomalies in the UI that the Windows Console (tm) version may not show, so you may find that the Windows Console (tm) version has fewer issues. The same code more or less is used in all three versions, so some of the differences are in the way that code is converted into the EXE file by the compiler, perhaps, or are in the OS or the way the compiler uses the OS, etc. The point being that one of the three versions might work on your Windows (tm) computer. A 32bit version of FreeDOS (tm) seems to be under development so the 32bit DOS version might run under that someday, perhaps. The GUI graphics viewer program DANGUIVU.EXE (tm) that should have been installed along with all the other files in DANCINEC.ZIP (tm) will not run on its own, if you try it will exit without doing anything. It is only able to show images made by DANCINED.EXE (tm), DANCINEC.EXE (tm). or DANCINEW.EXE (tm) at this time because those programs pre-process the image data through the needed LUT before passing it to DANGUIVU.EXE (tm) for display on your monitor. Press [t] to view the "Tutor" help screens in DANGUIVU.EXE (tm) and use the config setup in the calling program to setup the graphics window size and other passed config data values. Do not close the program that calls DANGUIVU.EXE (tm) before you exit it, as the calling program is meant to continue running after you quit DANGUIVU.EXE (tm). Note that if you use DANGUIVU.EXE (tm) as the frame processing test graphics preview viewer you should set the Windows (tm) desktop taskbar option "always on top" to disable, or if you are using DANGUIVU.EXE (tm) with a window size that overlaps the task bar on the bottom of the Windows (tm) desktop, Windows (tm) may send a window "re-draw window" command to DANGUIVU.EXE (tm) making it display the image twice when it opens, once behind the task bar and then again over the task bar. See the instructions in the Q and A section below in this document for how to disable the task bar "always on top" function in Windows (tm) to try to prevent the double drawing of the window on DANGUIVU.EXE (tm) program open. --- HARDWARE REQUIREMENTS MEMORY: 512MB minimum recommended, 2GB or more may be needed for processing 4K or larger files and on some tasks. Under Windows ME (tm) GUI you may need to manually edit some of the Windows (tm) system configuration files to get the computer to boot with more than 512MB in the motherboard, see the Q and A section of this file, and search the internet on how to deal with this known Windows (tm) issue. CPU: INTEL P4 (tm) type (or AMD (tm)) 2GHz or faster recommended, although slower computers could work for smaller image sizes or short projects. DISK: 500GB minimum recommended, 20TB or more may be needed for some projects, although you may be able to do the editing and processing one reel at a time (about 22 minutes of frames) so not all the disks would need to be plugged into a single computer at the same time, in fact several computers would be useful running in parallel to speed up processing of TB of image frames. Each time you double the K size of the frames you increase the processing time by at least four times. OS: Windows ME (tm) would be a good choice since DANCAD87.EXE (tm) and DANCINEC.EXE (tm) should run under it as well with all of the (VESA VBE 1.2, 2.0. 3.0 or DIRECT-X) graphics modes supported working. Windows ME (tm) needs the TB disks to be formatted FAT32, which is not a problem as there are freeware utilities to do that (such as fat32format.exe (tm)) and higher versions of Windows (tm) can read FAT32 disks up to 2TB and maybe more (4TB?). Windows XP home (tm) may work also and let you talk to NTFS formatted HDD, but there may be some issues with the VESA or DIRECT-X (tm) higher resolution graphics modes not working (in particular if you have use the composite video out on your video board in the past and then forgot to disable that or upgraded your video and the previous settings got "stuck" on with no way to disable them), and DANCAD87.EXE (tm) may not run well with some things like USB drives plugged in because Windows (tm) can crash or freeze so that even [Ctrl]+[Alt]+[Del] will not work. It is possible to reboot the computer with the FreeDOS (tm) CD (bootable CD) to run DANCAD87.EXE (tm) if your HDD are formatted FAT32 in order to avoid some of the Windows XP (tm) issues while you do color correction or sound mixing etc. You may also be able to reboot a XP home (tm) computer with the Windows ME (tm) rescue boot floppy if the computer's HDD are formatted FAT32 to avoid some of the later Windows (tm) version bugs when running DANCAD87.EXE (tm), but not DANCINEC.EXE (tm) as the (DRAM) system memory may not all be accessible under the ME (tm) rescue boot floppy for some reason, at least in its default state (that is you need ME (tm) GUI running for DANCINED.EXE (tm)). At the time I am writing this I have not tested the programs under Windows 7 (tm) but the Windows Console (tm) or Windows GUI (tm) versions may run under it, at least using the program DANGUIVU.EXE (tm) for the graphics frame preview, maybe. VIDEO: The 32bit DOS version (DANCINED.EXE (tm)) uses VESA video modes, not all current video boards support those, and they do not print which ones are supported on the box, so its hard to know what will work until you try it. The Windows Console (tm) version uses Direct-X (tm) but seems limited to 4:3 video modes between 640x480 and 1600x1200 for the most part, so if you have a 16:9 monitor you may need to plug in a 4:3 monitor to get the DIRECT-X (tm) full screen graphics preview to display. The graphics preview has a "probe" built in for measuring image tonal values to see if you have the grading right for your processed image (you press [P] after the test image displays) so there is some value in getting it working. Using the external viewer DANGUIVU.EXE (tm) should work on all versions of Windows (tm) from ME (tm) up maybe since it only uses very basic Windows GUI (tm) commands, and with all shape monitors, so you can try it if the other options for full screen don't work, see option #4 in the program's main menu, then option #2 for the graphics setup in DANCINEC.EXE (tm) and the others. --- PROGRAM USE Because the programs are designed for batch processing of Digital Cinema Camera image frame files, the image frame files need to be named using one of the supported file name formats. There are some rename commands in DANCAD87.EXE (tm) that may be of use to reformat the filenames, or you may be able to use the command prompt in Windows to rename files, or some small program may need to be written to do that task, you can let me know. If you are using the REDCINE-X (tm) program to make DPX or TIF frames, you should be able to edit the filename output string values prompt to get numbered frames that are eight digits plus the three letter extension, I call that "Padded" format and it is best for file import since Windows (tm) folder directory listings will sort the files in the right order. For the most part each numbered frame set to process should start with the frame numbered zero. The frames should count like, 00000000.DPX 00000001.DPX 00000002.DPX 00006789.DPX 00006790.DPX 00006791.DPX and so on. For RAW sensor data it would be, 00000000.RAW 00000001.RAW 00000002.RAW or 00000000.DNG 00000001.DNG 00000002.DNG and so on. The program(s) can also read "Long" filename type like, 0.RAW 1.RAW 2.RAW 6789.RAW 6790.RAW 6791.RAW But as mentioned Windows (tm) will not sort those frame files in the right order for a director listing when you look in a folder and want to find a particular frame to look at or copy a range of frames etc. The "Divided" filename format is used in my DANCAD87.EXE (tm) to overcome the speed issues with folders that hold more than about 500 files, so that real- time playback of frame files will work, and so that frame sets of two hours or more can be used. See the documentation about DANCAD87.EXE (tm) for more information on the "Divided" numbered filename format. In brief, the "Divided" filename format makes a set of sub-folders that have the name starting with the file extension of the file type being saved, each folder holding a maximum of 500 frames so 2000 frames would be in 4 sub-folders which are numbered as well. There is also a Divided type that uses 1000 frames per folder for projects that are longer, but use the 500 type if you can as it plays the PIX frames for editing with fewer glitches maybe (the program defaults to the 500 type). DANCAD87.EXE (tm) expects the PIX and other frames to be in the Divided filename numbering system, except maybe for image level I01 where Padded or Long could be used. Padded is sometimes also used for output of edited frames so other programs can read them all in order. When outputting a Padded frame set some OS may limit the maximum frame count to about 65532 per folder output to. TIF0000C/000.TIF TIF0000C/001.TIF TIF0000C/002.TIF TIF0000C/499.TIF TIF0000D/500.TIF TIF0000D/501.TIF TIF0000D/999.TIF TIF0001C/000.TIF TIF0001C/001.TIF If the file extension in Divided numbered type files is TIF then the sub- directories used will start with TIF, if KCC then KCC, if BMP then BMP and so on. With there 500 files per sub-directory Divided type selected the letters A, B, C, and D are used to break the frames into 500 frame groups to maintain access speed which can decline when there are more than about 511 files in a folder at one time, A and B are used for negative frame numbers (not supported in all functions) and C and D are used for positive frame numbers, C for frames whose last three digits are 000 to 499, and D for 500 to 999. After 999 the thousands carry to the sub-folder name's digits and the numbers in the new folders repeat the pattern 500 files C, 500 files D, 500 files C, 500 files D and so on. Before you begin to use the DANCINEC.EXE (tm) programs you need to select a camera type, if you are not using a RAW camera select the generic DNG camera type (from main menu 1 then 3 then 2). The DANCINEC.EXE (tm) programs are designed to be setup using the menu commands, and then you can let the program load the numbered source files and save the numbered result files, which could take many days for long frame sets of large high resolution images. Because each shot may require its own setup, the program can save each setup to its own *.GVN profile. To change the name of the current profile to save use main menu command #1 then #1. Be sure to change the *.GVN filename when you edit any of the setup values if you are not going to just execute the current settings. In the execute menu (3 from the main menu) you can select the command that makes *.BAT files, that can append the name of the current *.GVN file into its *.BAT code that runs under the Windows (or 32bit DOS) command prompt. The point of that is to collect many different sets of setups, and since the shot range for the source and result files is part of the setup, when you execute the *.BAT file from the Windows (tm) command prompt the DANCINEC.EXE (tm) program will be executed as many times as you appended and each time it will load its corresponding *.GVN file and so remember what settings to use for each range of shots to process. Such batch files could be setup to process many shots and so the computer would run itself from that "script" in the *.BAT file for weeks or months etc. There are some main steps to using the program to de-Bayer RAW camera data: 1) Select a camera type (if you are processing just RGB select Generic DNG). 2) Select a de-Bayer process method (or other process method, RGB to RGB etc.). 3) Select the source and result filenames. 4) Make a test frame and look at the result. 5) Execute the range of frames (and shots if more than one). Processing of RGB frames like DPX or TIF works the same except for step 2) where you would select a RGB process method. If DPX frames are going to be processed, then you need also to setup the DPX clip points and gamma adjustments so the input and output LUT can be setup right for the type of DPX (or CIN) files you will be working with. If you are only going to process RGB files, set the camera type to "Generic DNG". If you are going to export DNG data, you need to setup the *.DGF file with the right configuration for the data bits and black level and Tag 50721 matrix values. You will need to test the DNG files made to see if you have the setup done right as the header values can affect the color reproduction in some programs that read DNG files. Adobe (tm) supplies a "free" program you might download called something like DNG_validate.exe for checking that you have the DNG header values set to valid values (to some extent). If you are going to use the ADOBE (tm) RAW to DNG converter program to make DNG files to be read by DANCINEC.EXE (tm), be sure to go into the second level menu in that ADOBE (tm) program and select Bayer type DNG, not Linear DNG, as Linear DNG is already interpolated and that may degrade the results as well as prevent the file from loading as only True RAW Bayer type DNG files are supported, rather than make Linear DNG you should make 48bpp TIF or DPX files if you want to interpolate (de-Bayer) in some other program. --- BEFORE YOU BEGIN TO PROCESS FRAMES There are several things you need to setup before you begin to process frames with the DANCINEC.EXE (tm) programs. 1) Select a camera type, if you don't know or have a camera type select generic DNG. 2) Go into the graphics part of the program, command #4 from the main menu, and select the setup, then setup the video size your computer can use for the graphics preview, the default is 640x480 since most video boards support "safe mode" for Windows (tm) issues. In the Windows Console (tm) version the maximum resolution seems to be limited to 1600x1200 (you can try 1920x1440 and 2048x1556 if your graphics card and monitor support those, but if Direct-X (tm) and the compiler's library will work is unknown at this time as the highest I have seen working is 1600x1200 under Direct-X (tm) the 32bit DOS version does work up to 2048x1536 on some graphics cards and monitors), and as mentioned you may need a 4:3 monitor with 1600x1200 as its native resolution to get that to display full screen. 3) Go into the main menu option #5 sub-menu and make a project structure folder set. The "Structure" will be shared between DANCAD87.EXE (tm) and DANCINEC.EXE (tm). The "Structure" gives you a place to put the camera's RAW frames and to put the processed results, the PIX sync playback "proxy" frames used during editing (if you edit or mix with DANCAD87.EXE (tm)). It best to put the main project folder on the "C:" drive if you have enough space since the "C:" drive letter will not change as you add or subtract HDD from your system. Additional HDD can be used to increase frame storage space and you need to use the make "Structure" command on those drives to make a project folder with the same project number in order to share HDD space between several drive letters for the same project. All of the project folders do not need to be on the PC system at the same time, you and use removable drives, but the right drive needs to be in the system when you want to work with the shots and image levels that are needed. DANCAD87.EXE (tm) also has a make "Structure" command in its Kinema sub-menu, its the same code more or less and works the same way making the same kind of folder "Structure". 4) Because Windows (tm) can change HDD drive letters, if you have your projects files on many drive letters at the same time, and you add or remove a drive the drive letters can change, so you may need to re-assign parts of the filename strings in order to get the program to find the files (more so when using the Edit list made by DANCAD87.EXE (tm) while doing editing and mixing as the source of KCC filenames and such). So you can use the filename string replacement if you need to to untangle some of the problems caused by Windows (tm) changing the drive letters around on you, like if HDD J: becomes K:, you can enter search string as J: and replacement string as K: then the program will think HDD K: is J: and things should then end up on the right HDD, maybe. Be sure to put the colon in the search string since you don't want all the letters that match the drive letter replaced. See the "Utilities" sub-menu in DANCINEC.EXE (tm) for its file path string replacement command. 5) You need to setup the PIX frame size to make for use in DANCAD87.EXE (tm) as sync playback "proxy" frames if you want to use DANCAD87.EXE (tm) in that way, that command is in main menu option #1 then #5. If you do not have this setup with the right options the program will report errors that you are trying to fit images too large for the sync playback screen size, that is the PIX frames must be "windowed" to fit within the video resolution that will be used for playback, its possible for the image size or aspect to change during playback, but only if all the PIX frames fit the playback video screen size selected, but you cannot change the video bit depth, that is if you select 15bit video for playback, you cannot have some frames 8bit or 32bit as they will not look right, the computer cannot reset the video bit depth "on-the-fly" so all PIX frames must use the same bit depth, most of the time you would be using 16bit color mode for the playback frames, unless you are making a black and white movie, in which case you could use 8 bit mode to get a less grainy playback. 24 and 32 bit playback (32 is just 24 with 8 dummy bits but is done on some video boards that do not support 24 and can be faster in some ways, just not here) look better, but you need HDD or SSD that are twice as fast, or you need to reduce the playback image size to use those since the PIX frames would be twice as large as for 15 or 16 bit color images. --- STARTING TO PROCESS RAW DIGITAL CINEMA CAMERA FRAMES 1) Be sure you have selected the right camera company and camera model in the camera type menu (#1 then #3 from main menu in DANCINEC.EXE (tm)). There are options for X and Y mirror (flip) and negative image (such as from a film scanner shooting negative film images onto the sensor camera) in the camera model setup commands, so if the image is upside-down and such that is were you fix that issue, for manual setup of new camera types, you need to manually setup the Bayer order, there are four options to try until you get red, green, and blue parts of the image to look right (if faces are blue you have the wrong bayer order selected, for the know camera models that should work from the right model option for you already, unless the camera maker has changed something etc.) 2) If your camera's RAW data does not have a frame file header understood by my software, you can select the manual RAW camera type and enter the needed values. Some camera models that have a known maker, do not use a header on their RAW data, so you need to manually enter the image pixel size in X_pixels and Y_pixels, so for those cameras you need to note that when you shoot or use math and the file size to guess the right pixel dimensions. 3) The program is setup to process image frames by a "process method code number". The process methods are selected by using the menus in command option #2 from the main menu in DANCINEC.EXE (tm). Some methods do not have any adjustments, and other methods do, those sub-options are in menus that come up after you select the needed process method from the process method menus. There are also some header read and trace commands and other special utilities in the process method menus since they are associated with the process methods in or near those menu options and it seemed the best place to locate those seldom used commands, so you need to poke around in those menus to find a few "extras". The first process method needed is Engineering mode "Interpolated Color, no KCC" process method #20004 (Engineering mode sub-menu option #4). 4) Go into the execute menu option #3 from the main menu in DANCINEC.EXE (tm) and enter the source and result filenames and paths. You have several options for the source and result filenames: 1) Just process a single file between any source and result folder. 2) Process a set of numbered files between any source or result folder. 3) Process one shot in a source project folder and a result project folder. 4) Process a range of shot folders from the source project folder to the result project folder. Normally when processing frames in the project "Structure" the frames source and result Ixx folders are in the same shot Sxxxx folder, such as source RAW frames being in I01 folder and result TIF or DPX frames being in I04 folder. The "Structure" is meant to have 12 image resolutions/versions for each shot, you can use these as you like, but for some kind of default the programs think you are doing this, I01 = RAW or DNG source files, or ungraded TIF/DPX/CIN/BMP files I04 = graded or ungraded TIF files if the source data is RAW or DNG or graded TIF files if source is TIF/DPX/CIN/BMP. In the case of RAW or DNG source, the ungraded TIF could be overwritten when the grading of that shot is done so the TIF would be replaced with a graded one, if you don't want that you might put the ungraded TIF sample frame used for the KCC adjustments into I02 or some other Ixx level than I04, like I11 (you can have TIF, DPX, CIN, BMP, PIX all in the same Ixx level at the same time if you need to.) I11 = PIX "proxy" frames used for editing and sound mixing. Both DANCAD87.EXE (tm) and DANCINEC.EXE (tm) can make PIX frames, but DANCINEC.EXE (tm) may be much faster in making them, and has a "piggy-back" mode that lets it make PIX frames at the same time it is making graded TIF/DPX/CIN/BMP files to save time by not having to load and process the full size RAW or RGB frames twice (you add 1000 to the result type code in DANCINEC.EXE (tm) to activate "piggy-back" processing). So you can enter result type 1048 to get TIF (48) plus PIX (1000) result files at the same time. When you have more than one source or result file at the same time, the source or result filename option will repeat asking you the filename details the number of times, in this case the result would ask you for the result filename details twice since two result filenames are needed, which in most cases would just vary with the Ixx level and file type codes. 5) Once you have the source and result folders set and process method 20004 set, you can press [T] for Test in the execute menu (#3 from main menu in DANCINEC.EXE (tm)) to produce a test frame. The test frame is used for color correction adjustments in DANCAD87.EXE (tm) by using its color correction Grading command in the Files > Utilities > Kinema > Edit_list > [SpaceBar] > Grading command. The Grading command makes a special control file called a *.KCC file. To get DANCINEC.EXE (tm) to find the needed KCC file there are several ways to do that, gone into below, but for now you can just enter the name of the KCC made as the WBC KCC file in the #3 option command of the execute menu in DANCINEC.EXE (tm). You may need to use the #7 option in the execute sub-menu to turn on the graphics preview (DOS and Console versions only as of April 2011) if it is not on for [T] or Test execution mode. 6) Exit the execute menu back to DANCINEC.EXE (tm) main menu, then go back into the process method menu and select a workprint or finish process method. If there are presets for the finish mode for your camera you can select an option for your camera model from the presets menus. 7) Go back into the execute menu (#3 from DANCINEC.EXE (tm) main menu, and make sure the KCC files are set as needed using the #3 option there. 8) Make another test frame to look at in the graphics preview and or with some other graphics program by using the [T] or Test command in the execute menu as before in 5). 9) If the test frame looks good, you can then select option #8 in the execute menu to process the whole shot of frames, or a range of shots using the selected process method and color correction. 10) If you need each shot to have its own color correction, rather than use option #8 (execute), use option #9 to make a Batch file so that the process method, KCC files selected, and source and result files etc. can be unique to each shot or sub-range of shots. If you make a Batch file, i.e. *.BAT, you need to change the *.GVN file for each part of the Batch job in order have a new set of values saved and the previous ones kept for later use when the *.BAT file is executed from the Windows Command prompt. You can access the Windows Command prompt by using the Windows Run command in the Start button at the lower left, then enter the word COMMAND where you would type in the name of a program to run. Be sure to use the command line command CD to change to the directory that DANCINEC.EXE (tm) is in, and its also good to only save the *.BAT files into that folder so that the program and *.GVN files can be found. Some things to watch out for are: 1) Be sure that the process method used supports WBC KCC color correction to see what your KCC does to the image. 2) Be sure you use the prompt #3 in the execute sub-menu to enter the name of the KCC file made if you are using direct reference, or otherwise setup the KCC prompt for indirect reference or use the edit list sync options so the program can find the right KCC file to read. 3) Be sure that you use LEVELS#1 and CURVES#1 in DANCAD87.EXE (tm)'s Grading command rather than starting with LEVELS#2 and CURVES#2 as you can get very messed up results if you get that mixed up (which can happen if you don't read the prompts on the screen since it remembers the last one you used and when you go from Grading one shot to the next it will offer #2 when you want to start with #1 on a new Grading for the most part.). 4) The KCC names for the "presets" are named with extensions WBC, EDR, and SCP rather than KCC, although you can use them with the #3 option of the execute sub-menu if you want to manually select them, just remember to pick the right one for the WBC, EDR, or SCP KCC prompts. --- COLOR CORRECTION AND KCC FILES The color correction of both RAW sensor data, i.e. Bayer data in RAW or DNG files, or RGB image data like in BMP/CIN/DPX/TIF frame files, is controlled by something called a "*.KCC" file (Kinema Color Correction). KCC files are made by DANCAD87.EXE (tm) in its Grading command's control screens, Levels#1 Curves#1 Chroma_Matrix Curves#2 and Levels#2. (for Kinefinity.com (sm) KineRAW (tm) camera model type 504 Curves#2 and Levels#2 may be overridden by the camera's meta-data so in the WBC you would leave those at default for that instance.) In order to do "key-frame" color correction for each shot you first need to make a TIF Test frame from the source RAW/DNG data. For Bayer type sensors the TIF Test frame made by process method 20004 will look very dark and have a strong green cast and unsaturated color, that is normal it is what the "RAW" sensor data looks like (as well as being very soft focus because of the lens and the OLPF blurring the image). The dark and blurred look of True RAW data is normal and you will probably learn how to read it for checking highlight exposure and lens focus etc. in time and experience grading thousands of such un-corrected (non-white-balance) frames. Because processing Bayer sensor data well is a bit more complex than just processing RGB color separation digital scans from a film scanner where a monochrome digital camera is used to shoot three images through red, green and blue filters, three KCC files can be used to process each frame of Bayer Digital Cinema Camera data, one for the white balance and EI/ISO adjustments called the WBC KCC file, another to extend the highlight tones using some of the data cut off (clipped off) during the white balance, that is called the EDR KCC (Extended Dynamic Range), and a third used to make fine adjustments to the black and white levels, gamma, and color balance called the SCP KCC (Second Color Processing). To enable the presets and have all three kinds of KCC files in the same folder the WBC KCC is sometimes renamed *.WBC, the EDR KCC is sometimes renamed *.EDR, and the SCP KCC is sometimes renamed *.SCP, but those are for internal use by the program distribution, when you make KCC files (using DANCAD87.EXE (tm)) they would have the form *.KCC. When introduced for use only within DANCAD87.EXE (tm) for color correction of RGB to RGB files, the KCC files were always placed in the result files folder and found using the key-frame number for each shot in the edit list. For use with DANCINEC.EXE (tm) some other options for finding the KCC file needed to be introduced since DANCINEC.EXE (tm) can use three KCC files at once, and the presets do not associate the KCC files needed with any shot or key-frame so the KCC files can be in the DANCINEC.EXE (tm) folder all the time, or in other folders, or in the project's shot folders etc. This may see a bit complex, but there are reasons for having the KCC files used in more than one way, and DANCINEC.EXE (tm) is used both as a stand alone program for batch processing, and to process frames that will be used along with DANCAD87.EXE (tm) in the project folder(s) it is using. There are several ways to access KCC files: 1) Enter the KCC filename at one of the prompts in option #3 in the execute menu in DANCINEC.EXE (tm). 2) Enter a pointer to a KCC at one of the prompts in option #3 in the execute menu in DANCINEC.EXE (tm). 3) Use the Edit List sync options in the execute menu to have DANCINEC.EXE (tm) find the key-frame KCC file in the edit list. 4) Use presets (only works for cameras that I have sample data to calibrate with). 5) Have meta-data from camera select "look" matching presets (only supported in compatible cameras). Look for information on how those options 2) to 5) are used below. You can make your own *.GVN files that can act as a sort of "preset" since you can setup the KCC and processing options and then copy, clone, or re-use the same *.GVN except with the source and result filename options revised. If you have sample data from some additional Digital Cinema camera that there are no good presets for you can email me about that and with some sample data files from you I might be able to add more presets for everyone's convenience. --- WORKFLOW OPTIONS DANCINEC.EXE (tm) is designed to both work with my DI/NLE/MIX/CC system in DANCAD87.EXE (tm) and with other editing workflows. You can use DANCINEC.EXE (tm) just to de-Bayer Digital Cinema Camera data and output TIF/DPX/CIN/BMP frames that you can load into any program that can load such numbered image frame sets, such as VFX DPX workflow and CinemaDNG (tm) type DNG workflow. Likewise, you can just use DANCINEC.EXE (tm) to color correct TIF/DPX/CIN/BMP frames made by other de-Bayer software or converted from H.264, R3D, or other compressed camera formats, or from film scans saved (or made into) as RGB data. You can de-Bayer to BMP files then use freeware VirtualDub (tm) to make an un- compressed AVI file with sound if you have an AVI file. That can be used for a quick "workprint" resolution version of your footage to look at, or you can use "finish process method" to make a video like long AVI file of several shots to ingest into any editing program that can read un-compressed AVI files with the needed resolution and frame rate. When used with DANCAD87.EXE (tm) the Edit List sync options in DANCINEC.EXE (tm) let both DANCINEC.EXE (tm) and DANCAD87.EXE (tm) "talk" to each other through the current edit list settings, and so act together to enhance each others feature range. Just be sure that both programs don't use the Edit List file at the same time, or errors will result. For going out to other edit systems 16bpc DPX files may be your best option, that is selected in the result filename prompts in DANCINEC.EXE (tm). If the program you want to ingest your frames into does not support 16bpc DPX, try 16bpc TIF or if you need SMPTE time code try 10bpc DPX. If those don't work you can use fully color corrected BMP frames for export to other edit systems (gamma 2.22 BMP 24bpp), but you cannot re-grade BMP files without adding histogram gaps. BMP frame sets should only be used for making DVD and such after final grading has been done with higher bpc file types. With BMP output you have the option of using the CCIR/ITU 601 limits of 16-235 rather than 0- 255 for the tonal values, but you need to try and see if the downstream program will ALSO reduce the tonal range or not as you don't want it reduced twice (its best to reduce it in DANCINEC.EXE (tm) from the higher bit internal format since going from 8 bits to 8bits for the limits introduces scaling errors). If you find a file compatibility issue, please let me know and I then may ask you for sample files that work and don't work so I can see if there is some reason. Another option is to convert RAW Bayer data to DNG data, then pass those DNG files to a DNG based workflow such as CinemaDNG (tm). DANCINEC.EXE (tm) has options for bad pixel removal, so the bad pixels in the RAW camera data can be removed before the DNG file is saved, but the color quality of any DNG files made will depend on how you setup the black level tag, and the color matrix DNG Tag 50721. You would probably get better first pass results using the de-Bayer process methods under full manual color correction control to make TIF or DPX files and using those if you can, rather than go from RAW to DNG and then in another de-Bayer program DNG to TIF/DPX and then grade those results, but you can see what results you can get. Reverse de-Bayer to convert RGB frames to RAW or DNG is an experimental command set, and not advised for producing finished results by converting say DPX into DNG etc. The other way around is a far better workflow option. For workflows that need SMPTE time code you are best when using DPX to DPX for color correction, or RAW + meta-data to DPX for de-Bayer since other file formats do not have SMPTE time code to pass through. You may need to adjust the time code pass through options to "forward" the time code as needed in the result frame files. If your workflow do not support SMPTE time code, then you may be able to use the DANCINEC.EXE (tm) SMPTE time code burn in (window dub) options to burn in the SMPTE time code in the image area or letter box area etc. so you can see the SMPTE time code values ON your processed result frames. To summarize some of the workflow options: 1) De-Bayer RAW in DANCINEC.EXE (tm) and export 10bpc or 16bpc DPX frame files (DVD size, 1280x720, 1920x1080, 2K, 4K etc.), then load those into your current NLE workflow used for VFX work. 2) De-Bayer RAW in DANCINEC.EXE (tm) and export 48bpp TIF and 16bpp PIX frames to DANCAD87.EXE (tm) for DI/NLE/MIX/CC to produce a edited frame set and mixed down sound track as WAV files. DANCINEC.EXE (tm) can be used to add sub-titles and re-grade shots as needed. DANCINEC.EXE (tm) can also be used to do "opticals" which can be edited in using DANCAD87.EXE (tm) as additional shots. 3) De-Bayer RAW in DANCINEC.EXE (tm) and export 24bpp BMP graded frames, then use freeware VirtualDub (tm) to convert those and your camera's WAV files to un-compressed AVI or MPG files. Then Edit your AVI or MPG in your normal editing software as you have been doing. 4) De-Bayer R3D in REDCINE-X (tm) or some other program like VirtualDub (tm) and then use DANCINEC.EXE (tm) to grade those TIF/DPX/CIN/BMP image frames and export them to BMP/CIN/DPX/TIF and edit like workflows 1), 2), or 3) or in any other way. Or use DANCINEC.EXE (tm) to process those frames for "opticals" etc. 5) Use "freeware" called "5DtoRGB" (tm) to make DPX frame files from H.264 video files shot in HDSLR etc., then Grade those DPX frame files in DANCINEC.EXE (tm) and export graded TIF/DPX/CIN/BMP frames to the workflow options above or use DANCINEC.EXE (tm) to process those frames for "opticals" etc. --- COLOR CORRECTION STEPS FOR BAYER FILTER CAMERAS 1) Select the camera brand and model, adjust the image size, Bayer order, and XY flip as needed etc. 2) Use process method 20004 to make a un-corrected TIF file. 3) Put that TIF file in I04 level in a shot in the Edit Structure. 4) Use the Grading command in DANCAD87.EXE (tm) to open that TIF file with KCC at default settings. Set the result folder as I11 level to make the WBC KCC file. 5) Pick LEVELS#1 6) Set the RGB to all be highlighted, then adjust the black level. If the sensor had equal RGB gain, then the black level should be about equal for R, G, and B, maybe about 0.04 or something. 7) Leave the green white clip at 1.0, and reduce the red and blue white clip as needed to get the combined histogram to show the peaks aligned for a 90% white card in the frame. The IR filter cuts red, so under daylight the blue whit clip may be 0.6 and the Red white clip may be about 0.45. Under 3200K lighting, the blue white clip may be about 0.4 and the red white clip about 0.6 or something like that, it depends on the sensor and OLPF+IR/UV filter, the subject, and the lighting source type. Press the [*] as needed to re-draw the screen so you can see the adjustments you are making and how they impact the result. Don't set the black level to high to force the blacks to be dark, its better to lower the blacks with the S-curve later, when doing the Levels#1 its better to have some small space between the bottom of the image data and black clip so that you get balanced noise for the noise filters to work with in the de-Bayer processing (that is you keep both the positive and negative noise swing around the image brightness value, if you cut the bottom off you cannot recover the actual signal out of the noise since you have lost symmetry). 8) Adjust the gammas of the RGB to all be about 0.45. For very underexposed shots you can go lower, but don't go under about gamma 0.333 since you will get histogram gaps in the dark areas of the image, its better to stay close to 0.45 for normal to under exposure and adjust the Curves#1 later. For overexposed shots you can use gamma values between 1.0 and 0.45 but you might get more of a "video" look rather than a film like result. 9) If the mid-tone balance is a bit off, you can adjust the Red to Blue ratio (one up and the other down) until you get mid tone balance, the probe on the left side helps you pinpoint the mid-tone if you have something in the frame you can locate the probe over, the probe size can be adjusted with the num pad keys, and moved around on the image. 10) Exit the Levels#1 control screen and enter Curves#1 control screen. Set the curve controls to adjust all three colors (R, G, B). Push the shadow end of the curve down a little, and push the shoulder end of the highlights up a little, leaving the ends at full white and full black. Next adjust the center part of the curve to make a smooth "S-curve" shape. Press the [*] as needed to re-draw the screen so you can see the adjustments you are making and how they impact the result. 11) Adjust the probe and locate it over an 18% gray card in the shot if you have a test frame to grade from, otherwise locate the probe on the actors face or other important subject matter in the frame. Press [*] to redraw and see if the image tones are about 0.458 to 0.462 (0.18^0.45 = 0.462, 1/0.45 = 2.22 gamma correction) on the green in the probe readout. If the tones are to dark pull the center of the S-curve up a bit and redraw to check, or if too light you can pull them down a bit and redraw to check. When things look good, press [Q] to select the higher quality display, then press [Ctrl]+[C] to get the fill screen re-draw so you can look at the frame's color correction better. When you enter the Curves#1 command you are asked for the video resolution (1024x768 or 640x480) and the video color bit depth (15/16 or 24/32), the higher values will give a better looking full screen image, but as far as color correction goes, since the 15/16 bit modes are dithered the results are about the same for any of the video modes as you are looking for the overall color of the image not the color of fine details. 12) Exit the Curves#1 command, and enter the Color Matrix command. You need to increase the color saturation in the Color Matrix after you have white balanced Bayer sensor data using LEVELS#1. To do that, you increase (positive value) the value of the color that is the same as the color channel, and you decrease (negative value) the other two colors in that channel, which means that for Red Red is going to be a positive number and Blue Green are going to be negative numbers. How much the other two colors are negative depends on the intrinsic Bayer filter saturation on the sensor, you can tell how much saturation increase is needed by looking at the chroma vectorgram display in the upper left of the screen, if the chroma dots (blobs with noise mixed in) make it all the way near to the U shaped marks around the outside you have the chroma over- matrixed (natural colors are about 80% or less saturated so for normal colors on subjects you never get to 100% of the chroma vectorgraphic display range). Typical grading chroma matrix values to start with would go like this: Red part of Red output channel = 2.0 Green part of Red output channel = -0.5 Blue part of Red output channel = -0.5 Red part of Green output channel = -0.5 Green part of Green output channel = 2.0 Blue part of Green output channel = -0.5 Red part of Blue output channel = -0.5 Green part of Blue output channel = -0.5 Blue part of Blue output channel = 2.0 Note that within each output channel the total is ALWAYS 1.0, never have the total other than 1.0 as it will produce unusable images where the black and white points are not black or white. You will need to look at the chroma vector gram see how much chroma matrix color "gain" is needed for your sensor type. To get more orange skin tones, it is possible to adjust the chroma matrix so that you do not subtract green from red, or red from green, that pushes skin tones more to the yellow orange, but the exact values need to be adjusted by eye since various kinds of light (e.g. fluorescent), the IR filter, ND filter, and so can change the way skin tones look. An example of Red and Green having less subtraction might be: Red part of Red output channel = 1.5 Green part of Red output channel = 0.0 Blue part of Red output channel = -0.5 Red part of Green output channel = 0.0 Green part of Green output channel = 1.5 Blue part of Green output channel = -0.5 Red part of Blue output channel = -0.5 Green part of Blue output channel = 0.0 Blue part of Blue output channel = 1.5 Or Red part of Red output channel = 1.5 Green part of Red output channel = 0.0 Blue part of Red output channel = -0.5 Red part of Green output channel = 0.0 Green part of Green output channel = 1.5 Blue part of Green output channel = -0.5 Red part of Blue output channel = -0.5 Green part of Blue output channel = -0.5 Blue part of Blue output channel = 2.0 An example of chroma matrix for the monochrome EDR might be: Red part of Red output channel = 0.33 Green part of Red output channel = 0.34 Blue part of Red output channel = 0.33 Red part of Green output channel = 0.33 Green part of Green output channel = 0.34 Blue part of Green output channel = 0.33 Red part of Blue output channel = 0.33 Green part of Blue output channel = 0.34 Blue part of Blue output channel = 0.33 13) Exit back to the Edit List and re-enter the Grading command, but change result folder to I12 to make the EDR KCC file. Select "Clone" option and edit the KCC filename to Clone to I11, the KCC you just made for the WBC KCC file. 14) Since you cloned the WBC KCC file the settings in the EDR KCC file are the same, next you go into the Chroma Matrix, highlight all three colors and press [E] for "equal" matrix, that makes a monochrome image. 15) Exit Chroma Matrix and go into Levels#1, the highlight Blue and push the white point for Blue to 1.0, then highlight (with the right and left num-pad keys) the Red, then push the Red white point to 1.0. Use [*] to redraw the screen. If the mid-tones look dark, adjust the gamma for all three colors to make the mid-tone lighter in the monochrome result. You should now see a nice unclipped monochrome image, if the blacks look washed out you can adjust the black clip points for all three colors a bit (together if the sensor gain was equal to start with). 16) Quit DANCAD87.EXE (tm) and start DANCINEC.EXE (tm). 17) Go to the execute menu #3, and use its option #3 to enter the WBC and EDR KCC files you made. 18) Go back to the main menu in DANCINEC.EXE (tm) and select the finish process method 40001. 19) Go back to the execute menu and make a test frame to see what the result looks like. If any of the noise filters need adjustment go into the process method 40001 command setup and adjust as needed. If I have time later I may try to make a video of how to do those steps, if you get stuck you can try to email me and I will see if I can talk you through the process. Since the program is setup to work automatically from the GVN and KCC files etc. I might be able to make some for you if you send me a reference frame, we can talk about that if you need help. See also the page at www.DANCAD3D.com (tm) that shows the Grading command control screens, what I was talking about above should make more sense if you are looking at those screen captures. BTW you can capture those Grading control screens if you need to email someone what settings you are using, press the [~] key, a prompt will come up asking if you want to make a screen-shot BMP file, just use a short filename though since the space for typing the filename in there does not work with long filenames, just something like C:\LEVELS1A.BMP etc. --- DE-BAYER AGC ISSUES One problem with DNG based workflow and maybe RAW in general using other de- Bayer programs is that if the de-Bayer program automatically finds the clip points and adjusts the gamma or mid-tone brightness you might get a kind of AGC (Automatic Gain Control) effect, where as things move around in the frame the overall brightness of other parts of the frame can change even though they should not be changing. Many non-Cinema video type cameras have AGC to keep the subject contrast range within the limited range of their FM recording or compressed recording system such as H.264. That is very much a non-professional 35mm Cinema Movie camera result, in 35mm Cinema use the camera's exposure is mostly constant over the length of a shot, and the film-to-film printing also just selects a single "light" timing code for each shot during the grading of the footage for printing, the "light changes" take place on the frame line between shots, so you do not see them when the film is projected. DANCINEC.EXE (tm) is designed to work like a color motion picture printer where the RAW/DNG camera data come in, the color correction is adjusted for all the frames in each shot, and then the BMP/CIN/DPX/TIF graded frames are output all with the same corrections so that the brightness is stable over the length of the shot and you don't get the cheep camcorder AGC effect in your project. Note that de-Bayer programs that adjust to the light values in each frame may change the overall brightness of the image frames in the sequence acting like a kind of AGC, whereas DANCINEC.EXE (tm) was designed to lock the color correction for the length of the shot. If you want changes in the color correction of shots processed in DANCINEC.EXE (tm) mid-way through the shot you can use the lap dissolve process methods in the RGB to RGB "opticals" part of the process methods sub-menu. An example of where you would use this lap dissolve idea is where you pan the camera from shade to sun or the camera moves from inside to outside or outside to inside, you grade the shot twice once for the bright part and again for the dim part, then do the lap dissolve where you want to do the transition. In conjunction with doing the double grade and lap dissolve you would probably also adjust the iris on the camera's lens at the transition point in order to avoid over exposed highlights or under exposed shadows since if you bring the dark part of the shot up too much it will show more noise even if the image tones come out in the normal range. --- K VALUE AND OPTICAL FILTERS Kodak (tm) says that their clear noon "Daylight" value is 5500K and that standard "Tungsten" lighting is 3200K (or 3400K with some Photoflood (tm) lights). You will note that some Digital Cinema Cameras use 5600K for daylight balance, which seems to be some Video standard (maybe related to the CIE standards), that would seem to include more blue skylight. Other values like 6500K are also seen sometimes. If you set the Digital Cinema Camera to 5600K and then use an 80A filter, you end up 100K too blue. To have the Kodak Wratten (tm) filter system work the camera should be calibrated to 5500K. Camera lenses can have their own color bias with some being quite yellowish and others being more neutral, so the light on the camera's sensor may get filtered by the lens and so be at another K value than the light source on the subject photographed. It is false to think that you should not use optical filters any longer since Digital Cinema Cameras RAW data can be adjusted in post to get white balance no matter what the color balance of the light going into the camera was. The reason for that is that to get best results from the de-Bayer processing you want the Red and Blue signals to be equal and to shoot with the sensor at minimum gain and equal gain for all colors so that the noise level is not higher for the blue or red than the other. If you shoot under very low or high K value light you will get unequal noise in the Red and Blue signals because the clip points will need to be unequal and that acts as a digital gain making Red or Blue show higher noise levels. Also to get the best color results and fewer de-Bayer aliasing and moire effects its best to have the Red and Blue no less that about one f/ stop under the Green signal for neutral colored subjects and to have the Red and Blue signals about equal. Sometimes you want the results shot under "Skylight" to be a bit blue, and so you can not correct for the Red/Blue balance, but if you want natural neutral colors under "Skylight" you should shoot through at least a UV cut filter, or a light pink "Skylight 2B" filter. Even warmer filters like 81EF might be useful for very blue skylight before sunrise or after sunset (when there is no orange light from the sun on clouds or subject, just light from the clear blue sky). When shooting under "Fluorescent" lighting using an "FL-B" filter can decrease the yellow-green look that such lights can cause. Also UV cut (like Wratten 2E) can help since such lights can sometimes produce UV light that is not visible to your eyes but the sensor can pick up and unbalance the color in the recorded images. Using a (Red) "Enhancing" filter that has rare earth minerals in it can improve the brightness of red subjects which may look dark under Fluorescent lighting tubes as their Phosphor does not always include much Red light output since the Red phosphor is a higher cost one sometimes. When shooting outdoors the "base" EI/ISO of the camera for "Daylight" may be too fast for keeping the lens iris under f/8 (lens iris more OPEN than f/8 since stops between f/8 and f/32 will reduce the image resolution on small to medium size sensors). On small sensor Digital Cinema Cameras its is important not to stop the lens down too much as diffraction can reduce the fine detail in the image. To avoid stopping the lens down too much you need to use ND or other filters such as Linear polarizer. Normal dye based ND filters look gray to your eye, but can pass far red and IR light, if you use several stops of ND light cut, that can pass so much IR light that the IR cut filter in the Digital Cinema Camera no longer cuts enough IR light in ratio to the red, green and blue light needed for a color image to balance right, and the colors will get off balance and washed out. To prevent too much IR getting through the ND etc. filter pack in front of your camera, you need to add a extra IR cut filter. There are two kinds of IR cut filters, IR "hot mirror" type, and IR absorbing "green glass" type. The "hot mirror" type use interference layers and work well for "normal" and longer focal lengths, but not for wide angle lenses because the interference layers are light angle sensitive and so the corners of the image will pass more IR than the center of the frame when wide angle lenses are used. The "green glass" type do not cut the near IR light with a sharp curve and so also cut some Red light reducing the ISO and making the noise higher, but they do cut the far IR better maybe and work with any focal length lens if well made. So the "green glass" my require some more adjustment to the color balance when you process the frames. The "hot mirror" type can also cut some Red light but maybe less than the "green glass" type (IR) heat absorbing glass type. Another choice for ND light cut for shooting under bright lighting is a semi- mirror filter rather than a dye based gray filter. Such mirror filters are like a mirror but not 100% silvered so when you look through the back side you can see through. This type of glass is used in stores so people can watch for shoplifters through mirrors around the store. When you use semi-mirror filters the mirror part reflects both the IR+UV light and the visible Red, Green and Blue light in about equal amounts, so you do not get the visible to IR imbalance that normal dye based ND filters cause. You put the mirrored side facing the subject and other filters between the mirror filter and the lens, to reduce reflections. IR was not an issue with movie film cameras because the movie film was not sensitive to far IR and only special kinds made for day-for-night shots were sensitive to near IR light. Sensor cameras have very high IR sensitivity and all need an internal OLPF+IR/UV cut filter to operate well because the Red, Green, and Blue dyes in the Bayer filter on the sensor pass IR light. IR light is an issue not just because of the shift in the processed colors when there is IR mixed with the Red, Green, and Blue light, but also because many lenses do not focus IR light at the same point as the visable light so any IR light that gets to the sensor may be a bit out of focus and add blur to the processed images. For a "night" effect where the sky is black you can use a IR pass filter over the camera, but because there is an IR cut filter in the camera the light loss would be very high. IR pass filters look like a black glass, but if you look at a low wattage clear light bulb like a flashlight through them for a second or so you can see they do pass some deep red light (don't look for long as the heat may pass, and NEVER look at the sun through an IR pass filter as you may burn you eye and cause blindness). If you have many ND filters on the camera and point the camera at the sun you may still damage the sensor as some heat may pass through the ND filters even though there is little visable light, so you open the lens iris and let even more heat through to the sensor. UV was a big issue with movie film since most movie films would pick up UV light, although large movie lenses with thick yellowish glass would absorb some of the UV light. Sensors can also pick up some UV but the sensor is less sensitive to UV than IR so the need for UV cut may seem less that IR cut, but both are needed to get clean color reproduction and to compensate for the camera lens not being able to focus green and UV or IR light sharply with the focus set to the same distance. When you cut the IR/UV you decrease the "false" signals mixed with the Red, Green and Blue and get better results. You can see the color balance of your RAW exposures by looking at the Histograms in DANCAD87.EXE (tm)'s Levels#1 screen before you make any color corrections, if the peaks do not align, then you can see the ratio of the exposure as the scale is linear, half way over is about one f/ stop, one quarter as high is about two f/ stops down and so on. You can adjust the sensor's color balance with Wratten CC filters, such as CC20M (Magenta or Green cut) to reduce the green over exposure. CC and ND filters are rated by density, 0.3=1 f/ stop. 0.6=2 f/ stop, 0.9=3 f/ stop, 1.2=4 f/ stop, 1.5=5 f/ stop. CC filters come in Yellow, Cyan, Magenta so you can stack them to get shades of red, orange, green, and blue. They also come in Red, Green, and Blue but are less used that way since they do not stack as well as the Yellow, Cyan, and Magenta ones. The filters are marked like CC20M which means "Color Correction density 0.20 Magenta", the density number is adjusted to 100x so that there is no point, and the smallest values are 0.05 or 0.025 density or CC05 and CC025. Color values smaller than density 0.025 are hard to see with your eyes, but as motion picture print film was processed to about 2.5x to 3x (contrast near mid-tone) the contrast of your eye, using such minor filters in the printing light would produce a change that could be visible in the printed film made. On a sensor if you make the contrast higher in post, then you amplify the density ratio of any color filters used in front of the sensor or lens. To find the f/ stop change for the ND filter divide the filter density by 0.30, such as 1.8/0.30=6 f/ stops. Sometimes ND filters are marked 2X or 4X which may mean Density 0.3 or Density 0.6 etc. since 1 f/ stop requires twice the exposure and 2 f/ stops requires four times the exposure. When shooting outdoors using a Digital Cinema Camera you will need some ND filters since at 24fps and 180 degrees shutter angle, you have an exposure time of 1/48th second. The general rule is that in hazy sun the exposure is equal to f/11 if the shutter time equals the EI/ISO of the camera. But f/11 is too small a stop to use on small sensor cameras to get sharp results, and no Digital Cinema Camera has a native EI/ISO of 48 anyway, most are 320 or higher. Both of those factors push the camera to be very overexposed unless you use very dark ND filters for any shots made in direct mid-day sunlight. Even on overcast days or in the shade you may need some ND filter to keep the lens between f/6.3 and f/3.5 for sharpest results. A Linear Polarizer can act as a ND filter and when you rotate it you can also rotate it to control reflections that might be too bright or bring out colors by reducing reflections on surfaces. Using filters behind the lens can degrade the image because the lens may be only corrected to focus through air, which is why "gel" filters were used behind the lens, Wratten gel filters are very thin and so do not disturb the lens's optical correction as much as a thick glass filter would behind the lens. In front of the lens you can use glass, plastic, or gel filters for normal and wide angle lenses, but for telephoto lenses you should only use good quality glass filters with parallel faces other low quality filters can distort the light and blur the image. Sometimes because good glass filters are not at hand to use with telephoto lenses, gel filters will be used on the back of the lens where their lack of parallel faces is not so much of an issue. When you use a filter between the lens and the sensor, you need to compensate the optical distance by about 1/3 the thickness of the optical filter, or the focus marks on your lens will not align right and if you focus by tape you will get soft focus shots. Such compensation requires adjusting the flange-to-focal distance of the camera's lens mount (mostly). Zoom lenses need their optical distance right to track focus, so if you tape a gel filter on the back of a Zoom lens, you may need to compensate the focus marks a small bit (which could vary depending on the Zoom setting of the mm focal length). The primary complaint about using optical filters for white balance is that you lose some EI/ISO speed, that is you need more light or a faster lens. That is true, but if you do increase the light level AND use optical filters you will get better results than if you try to shoot under poor lighting conditions and with the light unbalanced on the sensor. There are times when you cannot do things "right" so you make the best of what you have, but that should not be your main plan for successful filmmaking. With Tungsten lights you can use Autotransformers or VariAC (tm) to increase the voltage going to the light bulbs and in doing so increase the K value and the brightness by as much as two f/ stops. When you "boost" the lights in that way, you can use a less dark bluish filter, so you gain two ways, 1) the light from the bulbs is both bluer and brighter, and 2) the filter needed is less blue and transmits more light so its loss is less. Therefore you can boost the lights and use a light blue filter and get full white balance at the same lens f/ stop as just shooting with no filter or lamp boost, but the end result will end up better since the blue part of the signal data with need less digital gain and so show lower noise, and the red pixels will not be overexposed and so with less clipping the overall color rendering should turn out better as well, and the results may look sharper since the green to red/blue balance will be optimized. No electronic or software method of white balance can produce results as good as using optical filters, if enough light is made available. I am bring this up in the DANCINEC.EXE (tm) documentation since one part of using the program is to adjust the "noise filters" which are needed to get a usable result from underexposed and unbalanced shots, well any shot but more so with degraded data. GIGO, Garbage in = Garbage out, although the software can improve the results quite a bit, you still get better results when you start with the best illuminated subject matter and the right optical filters being used. Here are some useful values for the K value for various kinds of light taken from a table in a Kodak (tm) filter guide and some from the American Cinematographer Manual: Skylight (clear, without direct sunlight) 12000K to 18000K (9500K to 30000K) Summer midday shade 7100K to 8000K Summer midday sunlight plus skylight 6500K Overcast Skylight 6000K Photographic Daylight 5500K High-intensity Carbon Arc light (sun arc) 5500K Sunlight (mean noon without skylight) 5400K White flame Carbon Arc light 5000K Morning or Afternoon sunlight 4300K Sunlight one hour after sunrise 3500K 500 watt "photoflood" bulb (34.0 lumens/watt) 3400K 500 watt 3200K photographic bulb (27.0 lumens/watt) 3200K 200 Watt general service bulb (20.0 lumens/watt) 2980K 100 watt general service bulb (17.5 lumens/watt) 2900K 75 watt general service bulb (15.4 lumens/watt) 2820K 40 watt general service bulb (11.8 lumens/watt) 2650K Acetylene (flame) 2415K Sunlight at Sunrise or Sunset 2000K Standard British candle (flame) 1930K Candle flame 1850K Match flame 1700K The "rule of thumb" is that 1 volt change for Tungsten 120 VAC bulbs changes the light 10K degrees, so if you need 200K more you would boost about 20 volts, but that is approximate. It goes the other way also, if you measure the voltage across your light bulb while on at the end of many long extension cords, you might see that your 120v source has dropped to 100v and so you have lost about 200K from your 3200K lights making them 3000K, so some CTB (bluish) gel filters would be needed over the lamps to compensate for the K loss. The K value for fluorescent lights is only approximate as they do not have a full spectrum like Tungsten light bulbs. When you shoot with lights you should use the same type for all the sources since you will otherwise get "mixed illumination" that can give off color shadows that cannot be compensated in full with color corrections. If you need to shoot with mixed light sources there are greenish gel filters you can use over Tungsten lights so that they better match fluorescent lights, and there are magenta gel filters you can use over fluorescent lights so they better match daylight or Tungsten lights. Here is some useful information about optical filters that could be useful with Digital Cinema Cameras: The blue 80 series filters make tungsten light closer to direct sunlight: 80D converts 4200K light to 5500K light, 1/3 T/ stop loss 80C converts 3800K light to 5500K light, 1 T/ stop loss 80B converts 3400K light to 5500K light, 1 2/3 T/ stop loss 80A converts 3200K light to 5500K light, 2 T/ stop loss If you light source is under 3200K then additional blue filters are needed: 82 converts 3100K to 3200K, 1/3 T/ stop loss 82A converts 3000K to 3200K, 1/2 T/ stop loss 82B converts 2900K to 3200K, 2/3 T/ stop loss 82C converts 2800K to 3200K, 2/3 T/ stop loss 82C + 82 converts 2720K to 3200K, 1 T/ stop loss 82C + 82A converts 2650K to 3200K, 1 T/ stop loss 82C + 82B converts 2570K to 3200K, 1 1/3 T/ stop loss 82C + 82C converts 2490 to 3200K, 1 1/2 T/ stop loss 78C converts 5360K to 5500K, ~1/3 T/ stop loss. 78B converts 5060K to 5500K, ~2/3 T/ stop loss. 78A converts 4660K to 5500K, ~1 1/2 T/ stop loss. 78AA converts 3460K to 5500K, ~2 1/3 T/ stop loss. 78 converts 2360K to 5500K, ~3 T/ stop loss. (Values for 78 series filters are approximate do tests before shooting.) So to convert 2650K to 5500K you could stack 80A + 82C + 82A filters, 3 T/ stop light loss. Light loss values are very approximate, check the camera histograms with an 18% Gray card or 90% white card to see compensation needed because the IR cut filter has a greenish cast and two filters of different colors can interact. Sometimes you need to shoot by open skylight without clouds or skylight before sunrise or after sunset which can be bluer than direct daylight so you want to "warm up" the light with amber colored filters: 81 converts 5600K to 5500K, 1/3 T/ stop loss. 81A converts 5700K to 5500K, 1/3 T/ stop loss. 81B converts 5800K to 5500K, 1/2 T/ stop loss. 81C converts 5900K to 5500K, 1/2 T/ stop loss. 81D converts 6000K to 5500K, 2/3 T/ stop loss. 81EF converts 6150K to 5500K, 2/3 T/ stop loss. 85C converts 7200K to 5500K, 1/3 T/ stop loss. 85 converts 7600K to 5500K, 2/3 T/ stop loss. 85B converts 7800K to 5500K, 2/3 T/ stop loss. 86 converts ~8640K to 5500K, ~1 T/ stop loss. (Values for 81/85/86 series filters are approximate do tests before shooting.) If you use optical filters on your camera to balance the light on the subject to 5500K, and you are using lights on the subject as well, then you need to use gel filters on your lights to match the light on the subject. Special large sheets of gel or plastic filter are made for such use, like CTB to convert tungsten lights to daylight or CTO to convert daylight to match tungsten. Those come in partial or whole conversion versions so you can get a little bluer or yellower light than standard, like full CTB, 1/2 CTB, 1/4 CTB etc. Those gel sheets are not made for use in front of the camera, but in a "pinch" when you don't have the right filter for your camera you may be able to cut out part of a sheet that does not have defects and use it in front of a normal or wide angle lens, or tape it over the sensor (watching out for dust). For longer focal length lenses putting a not optically perfect gel in front of the lens may give soft results, but putting it on the back between the lens rear and the sensor may work if there are no blemishes in the part you cut out. Under Fluorescent lights you can use a special magenta filter, FL-D converts Fluorescent to 5500K If you are using 3200K lights and shooting where there are Fluorescent lights then you need to do two things, 1) put an FL-D on the camera lens or in your matte box, 2) cover the tungsten lights with special greenish gel filters made by Lee filters (tm) or maybe some other brands. When shooting in daylight you will need ND (Neutral Density) filters because the "native" EI/ISO of todays Digital Cinema Cameras is too high to shoot at 24fps with the shutter set to 180 degrees without overexposure. To shoot in direct sunlight with small sensor cameras the EI/ISO needs to be reduced about ISO 8 to ISO 16 at highest since the lens iris needs to be more open than about f/6.3 to avoid resolution loss from diffraction. Wratten 96 0.10 density 1/3 T/ stop loss. Wratten 96 0.20 density 2/3 T/ stop loss. Wratten 96 0.30 density 1 T/ stop loss. Wratten 96 0.60 density 2 T/ stop loss. Wratten 96 0.90 density 3 T/ stop loss. Wratten 96 1.00 density 3 1/3 T/ stop loss. Wratten 96 1.20 density 4 T/ stop loss. Wratten 96 1.50 density 5 T/ stop loss. Wratten 96 1.80 density 6 T/ stop loss. Wratten 96 2.00 density 6 2/3 T/ stop loss. To set your light meter you divide the EI/ISO speed by the T/ stop loss, so if your camera is EI/ISO 200 and you add a ND 0.90 you get: 200/3 = 67 EI/ISO if you use a 1.80 density ND filter you get, 200/6 = 34 EI/ISO The smallest f/ stop you should use to get about 2K resolution is about: IMAX (tm) size sensor f/22 VistaVision (tm) size sensor f/16 S35 or A35 size sensor f/11 2/3" size sensor f/8 1/2" size sensor f/6.3 1/3" size sensor f/5.6 Peak resolution may be a stop or two more open than that for very good lenses, and a stop smaller or so for very bad lenses. For 4K resolution very few lenses are sharp enough no matter what f/ stop is used. If you stop the lens down more than recommended you will still get an image but the edges of the in focus parts will get softer, while the out of focus parts may get sharper. Sometimes it is better to get more of the frame is "semi-sharpness" than just the thin in-focus plane in very sharp focus. If your lens is marked in T/ stops you can stop down a little more because f/ stops are more open than T/ stop markings, like f/6.3 could be marked T/8. If your lens is not marked in T/ stops use this rule: Normal and tele lenses, open lens about 1/3 to 1/2 stop using the f/ scale for T/ stop target value. Retrofocus Wide angle lenses, open lens about 1/2 to 2/3 stop using f/ scale for T/ stop target value. Zoom lenses, open lens about 2/3 to 1 stop using f/ scale for T/ stop target value. The compensation between f/ stops and T/ stops is ALWAYS to open the f/ iris for a target T/ stop, say you want T/ 4.0 you would set the f/ ring to the f/ 3.5 mark on a normal lens, f/ 3.2 for a retrofocus wide angle lens, and f/ 2.8 for a zoom lens. The exact amount to open the lens depends on the number of lens elements, the color of the glass used, as some kinds are quite yellowish, and coating layers. Uncoated lenses lose about 5% of the light for each air to glass surface, single coated lenses maybe 2%, and multi-coated lenses might be about 1%, older uncoated lenses with 8 air to glass surfaces would then lose 5x8=40% of the light or about 1/2 stop. Zoom lenses can have 20 or more air to glass surfaces so 20x2=40% (1/2 stop) or 20x1=20% (1/4 stop), but you need to also consider the transmission losses in addition to the reflection losses so with more elements the total glass thickness the light goes through is greater. Some zoom lenses can show a lower T/ stop at longer focal lengths than their shorter focal lengths and so there might be a difference between the f/ stop and T/ stop of maybe one and a half to two stops at the long focal length end of the zoom range when the lens is wide open. The corners of the image are darker than the center when lenses are wide open and more uniform when the lens is stopped down, this difference can be almost a stop for lenses f/1.2 or faster. So when doing the EI/ISO conversion, take into account not only the filter's light loss, but the light loss of your lenses if they are not marked in T/ stops. --- OLPF ISSUES AND CHROMA In the finish de-Bayer process method 40001, and the finish RGB to RGB process method 95102, there are a number of filters to both blur and sharpen the image data. The Bayer filter is an imperfect idea that does not work 100%. Yet it is the way most color digital single sensor cameras work, why: 1) Its cheep since its just a monochrome sensor with a "checkerboard type" filter over it. 2) The bandwidth needed is 1/3 what a 3-chip RGB Raw camera would need for True RAW recording. 3) "Normal" lenses can be used with Bayer sensors, whereas only specially made lenses corrected for the glass prism used in 3-chip cameras can be used with 3- chip cameras (for the most part in the way things are done today). You can use a DOF-adapter with a 3-chip camera, but the light loss overall makes the system EI/ISO lower than using a Bayer filter sensor camera all things being equal with regard to the sensor. If you have a very sharp lens focusing an image on a Bayer filter sensor, and you point it at something like stars in the sky, the small points of white light will move about but the end result of the interpolated image will show the stars to be spots of bright colors that change as the camera is panned around. Why, because the sensor does not record color for every point as each pixel has a reddish, greenish, or bluish filter over it, so when a small white light moves around it is only seen as one or the other color but not as white if the lens is sharp enough to focus just on one pixel. That problem is called "Chroma Aliasing" and "Chroma Moire". To overcome the problem with Bayer sensors a special filter is used called an OLPF or Optical Low Pass Filter, which is designed to split the light rays going to the sensor into close spaced parallel beams that are spaced about the distance of two pixels so the sharp image is now a controlled blur on the sensor. To compensate for the blur on the sensor, the image is sharpened after de-Bayer interpolation, which un-does the OLPF blur, and if all that was done was that, you would not gain much other than extra noise in the image, since although the OLPF blurs the light image, it does not change the sensor noise, so when you sharpen to compensate for the OLPF blur you increase the image noise. To get ahead, the finish modes split the "luma" and "chroma" so that only the luma gets sharpened and the chroma is blurred a little more then processed with area filters that are more "intelligent" than a simple blur filter in that they do not "spill" the color outside the part of the image it should be in as much. The Bayer filters are not "saturated" filters that only transmit Red, Green, or Blue as would be used in a good 3 chip camera, rather they are just pale reddish, greenish, and bluish. That has two advantages and one disadvantage: 1) All the pixels record some detail information from subjects of any color, and when you sum the interpolated red, green, and blue images to get the luma image, like a monochrome photo, the detail mixes and so you get less "luma aliasing" from the fact that the response to light is uneven because of the Bayer filter. The weaker the Bayer filters are, the closer the "Luma" signal is in uniformity to a monochrome sensor. 2) The EI/ISO can be higher because the reddish, greenish, and bluish filters transmit more than just 33% of the light, improving the low light sensitivity of the camera. 3) Because the filters are not saturated, the reddish, greenish, and bluish image signals need to be processed through a chroma matrix to subtract the two other colors that "contaminate" the color you are trying to extract. Because green and blue do not exist where the red filter pixels are, and so on, you cannot get an exact matrix of the three colors as you could maybe in a 3-chip camera (equal balance for all pixels). When the reddish, greenish, and bluish images are interpolated there is uneven edge contrast do to 75% of the red and blue pixels being "filled in" with interpolated data that is somewhat in error. The 50% of the green pixels are "filled in" with inaccurate brightness data as well. The pattern of errors is uneven, and so that results in "Chroma Moire" where when you shoot a textured subject you see uneven and repeating patterns of hue errors in the finished image. That is because Bayer filter cameras are unable to produce accurate color for each pixel location because there is only one signal per pixel, not three for RGB at each pixel location. To overcome that issue two things are done, an OLPF filter is used in front of the Bayer sensor to blur the light so that the uneven edge contrast between the colors is reduced, reduced because the OLPF blurs the edges so there cannot be high edge contrast from the lens's image, and the second is that various blur and area filters are used to "even out" the processed chroma information so that when the color matrix is used to increase the saturation to compensate for the low saturation of the Bayer filters, you do not increase the chroma noise and Chroma Moire as much (that is the filtering the chroma of fine uneven texture before the chroma matrix reduces the chroma matrix from amplifying the chroma moire). After the Luma has been sharpened and the chroma has been blurred and area filtered then white balance matrixed the two are put back together to make a usable RGB image. As an additional step, after the white balance corrected image is made (WBC) it is possible to "overlay" and EDR or Extended Dynamic Range image over the white balanced image. When the Bayer data is white balanced the highlight end of the blue and red pixels is clipped off in order to have white be white and not pink or some other off color because the green pixels clip before the red and blue (the greenish Bayer filters seem to transmit more light than the reddish and bluish filters so green clips more than red and blue in the highlights of the RAW data), to compensate you clip the red and blue also in software. This red and blue clip is like adding gain to the red and blue to increase their contrast, but cuts off any signal above were the green pixels are clipped by the natural limits of the sensor's pixels and ADC range maximum. Because you need to clip the red and blue (but not the green) in the white balance step of luma and chroma processing, some highlight detail is cut off. The EDR step can put that highlight detail "back" into the image by making a monochrome image using non-white balanced information, that is you clip nothing in the highlights and use all the highlight detail recorded, then you mix the EDR monochrome image with the WBC white balanced image and increase the saturation of the result to compensate for the de-saturation of the mixing. The result of EDR processing with 20% to 80% mixing can be quite profound in that you gain quite a bit of highlight detail, maybe an additional f/ stop, and the images look less "digital" and more film-like since motion picture film is really made of 4 colors when EDTA bleach is used since not 100% of the silver comes out all the time so you end up with Yellow, Cyan, and Magenta dye images, and also some monochrome or brownish silver image (plus some stain from the development) combined. In the EDR step you can make the EDR overlay image a bit brownish or bluish or greenish etc. to simulate that effect, it does not need to be just neutral monochrome, such control is done with the EDR KCC file's color matrix adjustments that you do in DANCAD87.EXE (tm)'s Grading command. One side effect of the EDR is to add some chroma noise, so in the finish modes there are some area noise filters and active levels in the sharpen options to help clean up the blended image. As a final touch up of the color balance, you also have in the finish modes the SCP or Second Correction Pass KCC to make small color tint adjustments with the Levels#1 gammas, or to add an S-curve with the Curves#1 or Curves#2, or to make some color matrix adjustments. There are also re-size options in the finish modes to reduce and enlarge, as well as letter box or pillar box, convert to anamorphic etc. The reason I am talking here about the OLPF+IR/UV cut filters is that the many adjustments in the finish process methods relate to the OLPF+IR/UV cut filter as well as the sensor. You cannot make a good Digital Cinema Camera that does not have a good OLPF+IR/UV cut filter if it uses a Bayer filter. And even two cameras using the same sensor will give different results and require different finish process adjustments if they use different OLPF+IR/UV filters. The OLPF+IR/UV filter is almost as important as the sensor itself in what the end results will look like and how much aliasing and moire you will see when the camera or subject moves (you see aliasing runners more when high contrast edges move slowly over the Bayer filter "checker board".) So your understanding of why the many optical and software filters are needed is important to being able to make adjustments that improve the end results. They work together to do what the other cannot. There is no perfect software solution to replace the OLPF filter, and OLPF filters are not 100% effective, if they were then the resolution of the luma would be reduced, some detail needs to pass to the individual pixels in order to re-cover that detail with sharpen of the luma after the image has been blurred by the OLPF. Getting a usable image from a Bayer filter sensor is all about balance of the many factors in such a way that the contrast of the errors and artifacts is low enough as to not be obvious. If you boost the contrast, sharpness, and chroma saturation of the finished RGB de-Bayered images you can "bring back" some of the defects that were contrast reduced in the original de-Bayer processing. That is an important issue if you are going to Grade or re-Grade TIF/DPX/CIN/BMP files output from the de-Bayer, as you may reverse some of the "clean up" done during the de-Bayer steps to control the noise and artifacts. It can be better to do the final grading during the de-Bayer processing, rather than to grade or re-Grade the RGB files produced (these problems were not in 35mm film scans made with a monochrome digital camera and exposed as RGB color separation exposures since there was not Bayer filter involved that way). To do final grading in the de-Bayer as a "one pass" processing, you need to make a sensor "linear light" "test" frame and base the color correction on that frame, then go back and use that color correction template on all the other frames from the shot being color corrected, and in doing the corrections that way, no additional grading would be needed and therefore would not degrade the output of the de-Bayer processing. The issue of high chroma noise is not limited to Bayer filters though, color grains in color movie film to not overlap exactly, if you chroma matrix RGB scans of color movie film that were made with a sharp scanner, you will also "amplify" the chroma noise since the single pixel color information in movie film is also somewhat indefinite. To make the chroma more definite, you compare the color of the pixels around the one you are looking at, and in that way you get the chroma noise reduced. Such "look around" filters are called "area" filters in my programs menus. To process RGB film scans you can use process method 95102 as it has the extra area filters, if you just use the normal color correction on RGB images that have not been chroma noise reduced and apply inverse matrix values to increase the "purity" of the colors to make up for de-saturation in the film stocks, the grain will increase too much. Chroma low pass filtering is done in every color TV, DVD player, Blu-Ray player, Digital Broadcast, and VCR. What people think Super16 or 35mm film looks like is not what it looks like, what you have been viewing is very much noise filtered and chroma low pass filtered. Even in the old analog TV days 16mm films were projected onto camera tubes that had "stick" so that groups of three frames blended together in a kind of analog temporal noise reduction, and the color information was low pass filtered to the point that it was only about 70 lines of color information in the image width. For making feature films, you probably do not want to low pass the chroma as much as NTSC video was, but for people sitting a normal distance from a smallish screen in a "art house" movie theatre you can low pass the chroma to be about 5 pixels by 5 pixels in a 2K image, or 5/2000 = 1/400 the screen width. If you are shooting at low light and low K levels you may need to low pass the blue pixels even more since they would otherwise increase the overall noise in the image so you can go down to about 1/100 the screen width in the blue as peoples eyes do not resolve fine detail in blue but they will see blue spots or blue added luma noise. In an ideal world you would shoot with a 3-chip camera with large low noise sensors with saturated filters and optical white balance, so you might be able to get a clean usable RGB image close to a RGB film scan from IMAX (tm) size color negative, but we don't live in an ideal world, and sensors are not low noise enough yet to not need some kind of noise filtering in the de-Bayer or Color Correction. So you do what is needed to get a usable image from each cameras True RAW data. The various software filters in DANCINEC.EXE (tm) are adjustable so that you can use the minimum processing needed for each camera's data at the EI/ISO and K value used during exposure. --- TEMPORAL NOISE REDUCTION Temporal noise reduction is "stacking" frames to increase the exposure time for parts of the image that do not change much from one frame to another, and thereby reduce noise grain and increase visibility of fine detail in the processed images. Some advanced TEMPORAL noise reduction programs like NeatVideo (tm) or ARRI's Relativity (tm) use motion estimation to automatically track moving parts of the image to maximize the noise reduction. Although motion estimation might work somewhat for 60fps video images, 24fps images shot with 180 degree shutter have a great deal of motion blur, so from one frame to the next during rapid subject motion there is little "solid" edge on the subject to "track". If you stop the shutter angle down then you get more "stutter" to the motion at 24fps, or in the case of rolling shutter cameras more distinct skew as well. In the initial release of DANCINEC.EXE (tm) I have some Temporal noise reduction that works quite well for static camera shots to reduce noise overall in the non-moving parts of the frame where it tents to draw the eye away from the subject. There is not motion estimation used in my Temporal noise reduction, so there are no motion estimation related "artifacts" introduced to the shot made. The noise in the moving parts of the image is reduced by using the noise filters in the de-Bayer and the RGB to RGB process method that shares those (post interpolation) filter adjustments. The processing is very slow on about 2K images with a +/- 24frame window because of the huge number of pixels that need to be compared in order to get good results, so that is an issue, but if you can lock the camera down it can make shots that might otherwise be too grainy to be usable able to be used. Such shots would be those taken under low light levels, or when the sensor is underexposed more than normal in order to expand the dynamic range and hold as much highlight detail as possible. If the camera needs to pan its best to "whip pan" then freeze the camera motion since a slow steady pan rate for the most part blocks the frame mixing in my code at this point of development, depending on where you set the motion threshold, and if the threshold is set high then you would get extra motion artifacts (double images) which you would not want. You can run the shot through my Temporal noise reduction with the motion threshold just above the image's random noise level, the run it through some other noise reduction that has motion estimation like NeatVideo (tm) to maybe knock down the noise in the moving parts as well. You can also run the frames through the RGB to RGB process method that has the area filters to reduce noise all over the image. The trick in using noise reduction is to use the lowest settings that separate the image from the unwanted noise so that the impact on the image is to make the result looks better overall rather than worse or distracting. You can speed up the Temporal noise reduction by reducing the size of the moving "window" of frames checked, but the eye is very sensitive to small noise levels so to double the reduction you need to double the number of frames behind and in advance of the frame being processed, so you get a window of 3 frames, then 5 frames, then 9 frames, then 17 and so on so the jumps get bigger for about the same improvement and the time required grows by much greater degree. Another issue is for distribution of your movie on DVD, Blu-Ray, or Digital Broadcast, images with grain are not usable since there is not enough bandwidth to transmit grainy images through highly compressed formats. If you want your movie to be able to be used on todays compressed formats, and not just 35mm un- compressed film projection, then you must be mindful of what is needed to control noise grain in the finished results and avoid shooting in a way that would make it more difficult to de-noise your footage: 1) Shoot all shots with static camera as much as possible so that motion estimation errors from temporal noise reduction are avoided as much as possible. Motion estimation would be with software that has that feature. For software that lacks motion estimation, you need static camera in order for parts of the image to be static frame to frame so that noise reduction can take place at all. 2) For shots that cannot be shot static camera use reduced subject contrast ratio and follow camera along with maximum lighting and exposure. That is if you reduce the subject contrast ratio you can light higher before the highlights are lost, and in lighting higher the images would have less noise in the first place and so have less need for Temporal noise reduction. 3) Optimize grading of your shot so that you do not increase the image contrast, saturation or to sharpen images more than is needed, and any increase in contrast, saturation, or sharpness can make whatever noise there is in the images more visable. Its good to use the RGB to RGB process method that has the area filters if you need to re-Grade TIF/DPX/CIN/BMP frames since you can help control some of the noise build up that way (the area filters act within each frame rather than between frames in a sequence). 4) Use fill lighting or bounce boards when shooting so that you do not need to "Grade up" the mid-tones so you can see the actors faces as that just makes the images have higher noise levels. When shooting you should have an "eye-light" on the front of the camera's matte box to give a highlight in the actor's eyes which can let their face be a bit dark but still have the eyes resolved (it also makes it easier to focus on the eyes since you have a bright spot to focus on). An eye-light is a clear light bulb without a reflector so that it does not act as a bright fill light on the actor's face, it just adds sparkle to the actor's eyes as the light bulb filament reflects in the actor's eyes. The point being that if you want to use the temporal noise reduction in DANCINEC.EXE (tm) don't move the camera during the shot. Also shoot an additional second at the head and tail since the first frame to get maximum noise reduction with a +/- 24 frame window is one second from the ends of the shot. You should use the Temporal noise reduction BEFORE editing or doing the trim if you are going to extract frames from a shot because more frames are needed than will be extracted or trimmed out in order to get a uniform noise reduction effect as there is a small surge in noise at the head and tail of the processed shot because the window cannot reach past the frames it has to work with. See menus for Process Method 95710 in DANCINEC.EXE (tm). --- STATUS OF OPTICALS The basic "opticals" process methods should be working in DANCINEC.EXE (tm) such as fade, lap dissolve, matte composite, copy singe title frame into a shot, reverse motion etc. The RGB Key process method is mostly meant for super of title art where the background color is used to key a matte for putting a moving "plate" behind the title card art. The Chroma Key is very basic and needs about a two pixel encroachment to clean off any green band around the subject. Because the hard matte method does not work well with motion blur, you may want to reduce the subject motion and use small shutter angles to try to get a usable result. Also any reflections of the green screen off the sides of the subject would require outputting the chroma key matte alone, so you can re-touch the matte frames in a paint program then use the composite process method to combine the plate and foreground using that re-touched matte. Green screen is preferred to Blue screen since Bayer cameras have higher resolution in the Green 50% pixel fill, as opposed to 25% pixel fill in Red and Blue. You can get the keying reference RGB numbers for the green screen from the probe in the graphics test mode, you press [P] to have the probe appear when you see the graphics test view. (The graphics test view is in DANCINEC.EXE (tm)'s execute sub-menu #3 from its main menu.) When shooting for Chroma Key try to do these things: 1) Reduce subject motion per frame. Sometimes slow motion is used to reduce subject blur in high speed action films. You can also shoot at 48, 72 or 96 fps for printing down to 24fps which would give a shorter shutter time. 2) Reduce shutter angle. Sometimes small shutter angles are used to reduce subject blur in high speed action films. Small shutter produce an artifact called "picket fencing" or "strobe effect" with camera or subject motion, you can see this in movies such as "Gladiator (tm)". Its not so bad if the camera is at some distance from the actor so that his motions are a small distance in the images frame to frame to begin with. 3) Grease the subject's hair and use slick clothing like leather to decrease edge blur and get cleaner mattes. All three of those tricks seem to have been used on the "Matrix (tm)" movies. 4) Light the green screen with green gels over the lights to increase the saturation of the green screen. Try to use longer lenses and keep the actors as far as possible from the green screen so that the green light does not spill onto the actors front and sides. Hang a black drape with a hole to shoot through in front of the camera to avoid green light being reflected back onto the actor's front. Flag off the green gelled lights so that the do not shine onto the actors or sets. 5) Space the subject as far as you can from the green screen to avoid green light falling on the subject making it harder to separate the green screen from the subject. Giving the composite images an overall greenish tint helps hide any greenish cast caused by green screen light spill onto the actors, like seems to have been done in the Matrix (tm), i.e. use green screen for greenish tinted shots and blue screen for bluish tinted shots. 6) Use dull orange makeup on the actors to reduce the need to roto-scope out green reflections on their skin. You can re-grade the orange out as part of the composite matching to the plate. 7) Use smaller f/ numbers, but not too small as diffraction can reduce edge sharpness. Wider angle lenses may hold the edge sharpness better for some shots than very long lenses if there is more than one person in the shot. Focus as well as you can to get the sharpest edges you can as out-of-focus images would not work well with hard mattes. The Guide frame lines, SMPTE and other burn-in, and sub-titles is not an "optical" but done through the burn-in menu in the execute menu in DANCINEC.EXE (tm). If you find anything that does not seem to be working right let me know as some of these features are still in the experimental stage of development. Scrolling titles would be done using the main menu Preview command or a Macro in DANCAD87.EXE (tm) by offsetting a text vector element a little each frame drawn. The main menu can save the screen as BMP files, which should be good enough for white on black titles as grading will not be done on them, and so the 8bpc anti-aliasing should be enough, you can convert those to CIN/DPX/TIF in DANCINEC.EXE (tm) as needed. The anti-aliasing for the titles would make the images with oversample, you can also anti-alias by just making non-aliased titles that are 16K so in size, the use the re-size command to resample them smaller after doing a 3x3 or 5x5 pixel blur. Doing the blur then re-size down method may work faster for anti-aliasing titles. For the Subtitles in DANCINEC.EXE (tm) you can convert any of the DANCAD87.EXE (tm) vector *.FON files into the 64x112 pixel raster fonts, or use the vector font making commands in DANCAD87.EXE (tm) to design your own fonts, by using the font utilities in the Burn-in menu in the execute menu of DANCINEC.EXE (tm). Be sure to name the raster font *.814 in a way that you can tell what it is since there is no header to tell the program what the raster *.814 resolution is. In other words you need to be careful about selecting the *.814 files to use since you need to keep track of if their size matches the other settings for the burn-in and window-dub you select as a mismatch may result in a file load error or incorrect images of the symbols in the processed images. --- DNG ISSUES AND DNG/TIF TAGS DANCINEC.EXE (tm) can do several things with Bayer DNG (not so called Linear DNG which are like a non-white balance TIF file for the most part) files. Here is a list of some things DANCINEC.EXE (tm) is designed to do: 1) De-Bayer DNG files to BMP/CIN/DPX/TIF (this is the normal use). 2) Convert DNG to RAW without header for your special processing in some program you write or have. 3) Convert the RAW part of Bayer DNG to BIN for archive with lossless encoding to reduce disk space. 4) Convert RAW or BIN to Bayer DNG for processing in third party programs like DNG_validate.exe (tm). DNG_validate.exe (tm) is a "free" Adobe (tm) program that converts DNG files into TIF RGB files you can work with in graphics programs. 5) Convert BMP/CIN/DPX/TIF to Bayer DNG by using "reverse de-Bayer" (this is experimental). 6) Convert DNG to DNG so that adjustments can be made to the TIF/DNG Tags in the header. 7) Trace the DNG header to a TXT file so you can read some of the settings used. 8) Adjust the DNG header processing and values used in its DNG configuration. If you are going to make DNG files from RAW files, there are several settings you need to adjust for the DNG header tags that have to do with the sensor and color balance in order for the downstream program to get the color correction close to usable maybe. Many of the TIF/DNG tags are more or less straight forward, and you can see how those are used for any given camera's DNG files by using the DNG header trace. Some of the TIF/DNG tags are less easy to understand, but still need to be adjusted right for the sensor and OLPF+IR/UV filter used in the camera or camera's type of DNG file you want to make. In TIF/DNG Tag 50778 if you see a code number over 32768 it means that its a light color temperature K value plus 32768, for example in some Acam dII (tm) sample files you may see a value of 38268 used with Tag 50778, you can convert that to the K value like this, 38268 - 32768 = 5500K. The TIF/DNG Tag (put there) for black level also needs to be adjusted since RAW sensor data should not have the black level set to zero data value, it should be closer to a value of like 0.04 of full scale like, 12bit is 0-4095 so 0.04*4095=164 with some noise the sensor data with the lens capped would then be about 160 to 168 except for "hot pixels" and other defects like fixed pattern noise. The TIF/DNG Tag 50721 controls the mapping of the camera's sensor data's color gambit to the end RGB result images in programs that read DNG files and use Tag 50721. So when you make DNG files to be used with your camera's data, you need to match the TIF/DNG Tag 50721 color matrix XYZ to RGB values that go with your camera, or the results will not show close to the right colors. The Adobe (tm) color model used with TIF/DNG Tag 50721 seems to have been developed for use with the still image cameras and four color printing industry, rather than being optimized for motion picture color grading and de- Bayer processing of small sensor cameras, and so may result is less than best quality results as relates to how the third party de-Bayer program interprets the Tag 50721 values, and if they were set "right" for the color of the light and sensor type etc. in the first place. I communicated with someone at Adobe (tm) (Adobe (tm) is in charge of the TIF/DNG Tag oversight) about if they have a pdf document on how to figure out the right XYZ values for Tag 50721 for a camera where the XYZ values are unknown and as far as I can make out no such document exists (as of April 2011). So it seems the only way for you to figure out the right values to use for a new sensor or OLPF+IR/UV type is to guess and the process a color test chart and measure the colors you get in the finished RGB image file, then readjust the Tag 50721 values and try again, until you get as close as you can to some kind of usable output, maybe. Some discussion on CML seems to have leaned in the direction that having the de-Bayer depend exclusively on Tag 50721 should not be expected to yield finished quality color correction ready for your final edit, and that the choice is to use a de-Bayer program that ignores Tag 50721 (as my programs do currently) or to try to re-Grade the de-Bayered RGB frames in some color correction software. With my DANCINEC.EXE (tm) I don't use Tag 50721 except for writing DNG files since it is mandatory for other programs that will read the DNG made, rather I use the KCC files made in DANCAD87.EXE (tm)'s Grading based on a test frame to do the final color correction as part of the de-Bayer in a single pass. So you should not be surprised if the results you get when you de-Bayer DNG output from DANCINEC.EXE (tm) do not match the de-Bayer results made with the process methods that de-Bayer in DANCINEC.EXE (tm). It seems some commercial software also has an option to ignore Tag 50721 in its de-Bayer processing and rather uses its own color correction settings. Note that even if two cameras use the exact same sensor type, the nine XYZ values for DNG Tag 50721 needed may be different because of the color bias of the OLPF+IR/UV cut filter and other adjustments of the sensor black level, gain and such. To aid you in matching the Tag 50721 "look" used for each camera brand, here are the Tag 50721 values used in some DNG files from their cameras. Note that the Tag 50721 values are encoded in the DNG files as two numbers in a fraction, but I show them here as decimal values. To convert the decimal values to the fraction do this, Numerator = (decimal * 4095) Denominator = 4095 You can use 10000 as well, but some cameras may need values that would overflow the 16bit range of the Numerator, so using 4095 should be accurate enough for most uses. The order of the XYZ values may change with the Bayer order, so refer to the Bayer order for each camera or sensor to see how it relates to the nine matrix values. Canon XTi (tm): Camera set to Daylight (as far as I can recall, image was shot on scanner through CC filters) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 0 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 1 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 1 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 2 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 0 TAG 50778 Calibration Illuminant1 = tag not used in this Canon XTi (tm) file. TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 0.7622 -0.2223 -0.064 Color Matrix values 4 to 6...: -0.6442 1.3441 0.3359 Color Matrix values 7 to 9...: -0.0678 0.0955 0.7621 Kinor-2K (tm): Solders standing near stone buildings. From file: Kr19_01_05.0000002451.dng (rename 30100001.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 64 TAG 50778 Calibration Illuminant1 = tag not used in this Kinor-2K file. TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 3.240479 -1.53715 -0.498535 Color Matrix values 4 to 6...: -0.969256 1.875992 0.041556 Color Matrix values 7 to 9...: 0.055648 -0.204043 1.057311 Green screen shot of woman with playing cards. From file: g081.0000000266_1.dng (rename 30100002.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 0 TAG 50778 Calibration Illuminant1 = tag not used in this Kinor-2K file. TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 3.240479 -1.53715 -0.498535 Color Matrix values 4 to 6...: -0.969256 1.875992 0.041556 Color Matrix values 7 to 9...: 0.055648 -0.204043 1.057311 Acam dII (tm): Light House image sample. From file: Nick Paton-A-cam dII-01100015.dng (rename 40100001.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 122 TAG 50778 Calibration Illuminant1 = 38268 or 5500K, [(n-32768)=K] TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 1.0048 -0.2724 -0.05 Color Matrix values 4 to 6...: -0.5656 1.3498 0.2078 Color Matrix values 7 to 9...: -0.2916 0.3498 0.6394 CU cool light guy with glasses. From file: 01809356.dng (rename 40100002.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 0 TAG 50778 Calibration Illuminant1 = 38268 or 5500K, [(n-32768)=K] TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 1.0048 -0.2724 -0.05 Color Matrix values 4 to 6...: -0.5656 1.3498 0.2078 Color Matrix values 7 to 9...: -0.2916 0.3498 0.6394 CU couple on Bus (?) at night. From file: 20000475.dng (rename 40100003.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 0 TAG 50778 Calibration Illuminant1 = 38268 or 5500K, [(n-32768)=K] TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 1.0048 -0.2724 -0.05 Color Matrix values 4 to 6...: -0.5656 1.3498 0.2078 Color Matrix values 7 to 9...: -0.2916 0.3498 0.6394 Lake view Daylight. From file: 02015450.dng (rename 40100004.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 122 TAG 50778 Calibration Illuminant1 = 38268 or 5500K, [(n-32768)=K] TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 1.0048 -0.2724 -0.05 Color Matrix values 4 to 6...: -0.5656 1.3498 0.2078 Color Matrix values 7 to 9...: -0.2916 0.3498 0.6394 Steenbeck (16mm editing flat bed table). From file: 01100100.dng (rename 40100005.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 1 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 0 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 2 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 1 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = 122 TAG 50778 Calibration Illuminant1 = 36168 or 3400K, [(n-32768)=K] TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 1.0048 -0.2724 -0.05 Color Matrix values 4 to 6...: -0.5656 1.3498 0.2078 Color Matrix values 7 to 9...: -0.2916 0.3498 0.6394 SI-2K (tm): From file: sbodies_nov20_20_Nov_2009_13_55_32.dng (rename 60100001.DNG for my preset) TAG 33422 Bayer order: [TAG 33422] Lo-Lo byte.......: 0 [R=0, G=1, B=2 UL (shr 0)] [TAG 33422] Hi-Lo byte.......: 1 [R=0, G=1, B=2 UR (shr 8)] [TAG 33422] Lo-Hi byte.......: 1 [R=0, G=1, B=2 LL (shr 16)] [TAG 33422] Hi-Hi byte.......: 2 [R=0, G=1, B=2 LR (shr 24)] TAG 50714 Black Level = (tag not used in this file) TAG 50778 Calibration Illuminant1 = (tag not used in this file) TAG 50721 Color Matrix: Color Matrix values 1 to 3...: 3.24047899933 -1.53714999981 -0.49853499986 Color Matrix values 4 to 6...: -0.96925599946 1.87599099884 0.04155599976 Color Matrix values 7 to 9...: 0.0055647995 -0.20404299933 1.05731100006 To configure the DNG output for a new sensor type or if you change the OLPF+IR/UV cut filter you need to go into the DNG special processing menu's sub-menu and select the option marked "90150" then use the additional options to setup the *.DGN file that holds the values for the tags shown above so that the values you enter will be written to the DNG files when you use a process method that will process image data in the execute menu (execute menu is option #3 in the DANCINEC.EXE (tm) main menu, the DNG setup is in the process method menu which is option #2 in the DANCINEC.EXE (tm) main menu). See option 95155 (its not a process method, but its in the process method menus because it relates to the process method setup for output of DNG frames). As was mentioned the Tag 50721 values are stored as a fraction, so you can relate the decimal XYZ to RGB Color Matrix values show here by way of decimal to fraction conversion, like 3.2404 can become 32404/10000 etc. You should be able to use my DNG header trace command to see the XYZ values as the fraction's values or decimal if you want to see how each camera maker set the exact values used in Tag 50721 for their DNG frames. --- DARK FIELD SUBTRACTION AND BAD PIXEL REMOVAL Solid state sensors such as CCD and CMOS are not perfect and have some pixels that are "burned out" and can show up as black or colored spots on the end results. Sometimes whole rows or columns of pixels might be defective making a line in the image. In cameras that have lossy compression, such defects need to be "removed" from the sensor's RAW data by interpolating over them before the shot is recorded in the lossy compressed format (you would not do the removal after de-compression since the pixels would be blurred together and you could not single out just the bad ones to fix). True RAW recording Digital Cinema cameras can record the images with the "bad pixels" data intact, and later remove the "bad pixels" and "fixed pattern noise" (FPN) as part of the de-Bayer (de-Mosaic) processing. Also with True RAW Digital Cinema Cameras you can record the "black frame" data (lens capped) before or after shooting the shot you need since the shot is or can be recorded saving everything that the sensor outputs both good and bad intact. There is some advantage to being able to record the "black frame" set used to automatically find the bad pixels with, since if a pixel goes bad in the middle of a shot if you record the "black frames" before you shoot the shot, the new "bad pixel" will not be marked as bad, but if you record the "black frames" after the shot is made you will probably also be recording the new "bad pixel" as well (and so it could be fixed automatically). In some lossy compressed cameras if a pixel goes bad after you "black calibrate" then the bad pixel would maybe be recorded in some blurred state along with the other data that gets effected by the lossy codec. That is such cameras need to be re-calibrated after a new pixel goes bad, and then the shot would need to be re-done or somehow re-touched since the lossy encoding cannot be reversed 100% back to the original sensor data to interpolate just the area of that single pixel that went bad during the shot made. If you have a True RAW Digital Cinema Camera that does not do its own bad pixel removal in the camera or its download software, I have command options built into DANCINEC.EXE (tm) to do the needed bad pixel removal. You should remove the bad pixels when you are going to convert RAW data to DNG files since the DNG specification calls for the bad pixels to be removed before the RAW data is saved into the DNG file as that helps speed up automatic processing of the DNG frames and such. DANCINEC.EXE (tm) has three parts to its "bad pixel" removal: 1) Interpolate over pixels that exceed a "dark field" brightness tolerance. If you set the tolerance too tight many pixels will be take out of the image, so you just want to "remove" the pixels that are very much out of tolerance. 2) "Dark field" subtraction of Fixed Pattern Noise (FPN), in CMOS sensors there can be some vertical lines in the image that can be seen in dark areas and areas of low detail such a the sky. Since the Fixed Pattern Noise has a signal level about the same as the random and moving pixel noise, you need to shoot several seconds of "black frames" in order to get a "clean" image of "just" the Fixed Pattern Noise. The program has commands to do that for you, it combines many frames into a single "dark field" reference frame for your camera's sensor. If you have more than one camera you will need to setup more than one program folder for each camera or otherwise keep track of what "bad pixel" files go with which camera. You cannot use a "dark field" made from one camera's data on another camera's data, if you do then you will be ADDING fixed pattern noise and not be "removing" it. The black level can drift in some cameras so you should do the shooting of the "black frames" after the camera has warmed up to the temperature it would be when you are going to be shooting. Since there can be some drift in the black level, and new bad pixels can form, it would be best to record 15 seconds of black frames at the start of each days shooting, and another 15 to 30 seconds at the end of each SSD/HDD worth of frames. 3) "Dust field" subtraction is in the camera as an additional form of FPN solution, but you would need to shoot dust field frames by shooting with a translucent white plastic sheet or lens cap on the camera in order to try to get an image of the dust that matches what you will get when you shoot or shot the subject. Although it might be a good idea to shoot such "dust field" frames for an emergency use before or after each shot, how well they might work to save a shot that would otherwise be unusable needs to be looked into some more. I would not depend on the "dust field" subtraction to work in any useful way, its something I added to see if it could be of some help in avoiding a total loss on shots where there might be some dust in the camera that goes unnoticed at the time of shooting. Also there are times where taking the lens off the camera would let so much additional dust into the camera, that you might need to shoot "as is" and try to fix the dust in post later using a set of "dust field" frames shot at the time of the shot you want to try to "repair". In addition to the FPN are random sensor noise, a bit like video noise or film grain, and moving pattern noise which is probably caused by digital noise from the camera circuits (sometimes inside the sensor itself) getting into the pixel bias, pixel pre-amps, or causing small timing errors in the pixel addressing etc. Moving pattern noise looks like a kind of flicker but tends to be in bands that flash or roll over the images when you let the frames playback at sync speed, it is harder to see looking at a single frame. There are filter options in my de-Bayer finish process method 40001 to reduce random sensor+preamp+ADC noise and make the images less "grainy" by not sharpening the whole frame but sharpening sharp details more that areas of low contrast. Those filters work on each frame just within that frame. Another way to reduce "random" grain noise it to use Temporal Noise Reduction which "stacks" many images to average out random noise to levels lower than within just one frame. I have Temporal Noise Reduction in a preliminary state in DANCINEC.EXE (tm) that can work with static camera shots. Other Temporal Noise Reduction programs can be used that have "motion estimation" to deal more or less well with moving camera shots or rapid subject motion, which is not easy with 24fps footage do to the higher motion blur than found in 60p shooting. Moving pattern noise "dances" around in the frame area, here in one frame and there in another frame, so Temporal Noise Reduction can do a good job of reducing it in parts of the image where there is little motion, like walls etc. in static shots. In parts of the image where the subject is moving you do not see the moving pattern noise as much because the contrast change is much higher where the subject is in motion in the frame. The net effect of using all of the bad pixel removal and noise filters etc. is to transform the almost useless image quality you see if you look at the RAW data un-processed into finished color corrected images that are quite close to Super16mm or even 35mm movie film, and in some ways better since you don't have the serious dust, scratch and muck problems that working with film generations introduce. Today, film scans made for doing DI are also processed through "dust busting" software and area filtered to "de-grain" them, if that was not done film would not look as good as people thing it does, what people thing of as the "film look" today is a hybrid of older film technology and newer digital image processing along with filtering for compressed broadcast or DVD and Blu- Ray disk production where the images must be de-grained in order to reduce the bandwidth required for fitting the movie on one disk and such. I have made some shortcut commands for doing the "dark field" and "bad pixel" setup in the execute sub-menu #3 from DANCINEC.EXE (tm)'s main menu. And you have more full access to the "dark field" and "bad pixel" commands in the setup sub-menu #1 from DANCINEC.EXE (tm)'s main menu. The "dark field" and "bad pixel" commands make some reference files, if you have a project structure made, you can put the "black frames" in a shot and then use the shortcut commands to make the reference files in there and those commands set the filenames for the reference files in the GVN file so use when you process your RAW camera data, so that the "clean up" of the camera data goes automatically as far as the bad pixel removal and dark field subtraction. The reference files are the TIF "dark field" image which looks almost black, to see what is in it you need to make a color corrected version that has the black and white clip points set very close together around the sharp histogram peak. The "bad pixel" image can be a BMP file, you can manually mark extra pixels bad if you need to by using a program like "MSPAINT.EXE (tm)" setting the pixels red for "hot" and blue for "cold", hot being over black level tolerance range and cold being under black level tolerance. (Most of the bad pixels seem to be in the over tolerance group on the sensors I have tested data from, and so would be marked red.) To get best results from the sensor the black level should NOT be set to 0 signal, rather it should be up around 0.04 of signal 0.0 to 1.0, in the case of 12 bit data being 0 to 4095, 0.04*4095 = 164. The reason for that is that you want both the high and low peaks of the random and fixed pattern noise in order to filter them out, if you only have the positive peaks you cannot "balance out" the true center of the black level. After dark field subtraction the black level would be centered on 0.0 rather than its original 0.04, so there is a "black level bias" you need to adjust for your sensor type in the "dark field" setup command. You want the "black level bias" to be just a little less than the center of the gauss peak from the histogram of the original un-subtracted RAW camera data in order not to clip any of the highlight detail, if your original data was black center at 0.04 and you re-bias the "dark field" subtraction of your RAW data at 0.05 then you will cut off some of the highlight detail, so you would want to use a value of 0.04 or a bit lower like 0.039 for the black bias after dark field subtraction. You can use the histograms in DANCAD87.EXE (tm)'s Edit list Grading command to find the center of the "dark field" file's black level, that is the "dark field" TIF file made buy the "dark field" and "bad pixel" shortcut in the execute sub-menu in DANCINEC.EXE (tm). Each camera's RAW data will have its own width to the noise peak and its own center value for that noise peak in the "dark field"'s histogram. The "dark field" histogram is (can be) like that of a monochrome sensor since there is no light on the sensor when the data is recorded, and the sensor gain for a Digital Cinema Camera would be set to equal and minimum for all four colors, red, green one, green two, and blue. You can look at the histogram for the "dark filed" by de-Bayering to a TIF file without color correction, then viewing in DANCAD87.EXE (tm)'s Grading command in its LEVELS#1 control screen, you set the adjustments to work on all there colors then move the black and white clip points very close to the peak you see in the three input histograms, that will increase the width of the output histogram so you can see its "bell curve" shape for the noise distribution. The black level may drift as the camera gets hotter or colder, there are two ways to deal with that: 1) Adjust the dark field black bias value up or down a bit. One reason that the default bias should be a bit lower than the sensor average black, since you cannot increase the black bias without clipping the highlights otherwise. The black level must be stable since all the color correction KCC files are based on a single black level for the image data. 2) Process another set of dark field (black) frames that would have another black level up or down from the last set processed, that is you update the "dark field" TIF and "bad pixel" templates to values closer to those the sensor had when shooting the shots you want too process. If you shoot black frames at the start and end of each "digital magazine" your camera takes, then you will have the option later of making revised reference frames to process your shots with. If you do not have any black frames from that days shooting, then you are left with adjusting the black level bias value a little and manually editing the bad pixel template if needed because some pixels when bad on that days shooting. So I have ways to deal with some of the sensor issues, its easier to do automatic processing from recent black frames shot close to (and after) the time the shots were made, but if you were not able to do that because you ran out of storage, you may be able to "fudge" the program into compensating for any extra noise caused in the graded results by a high temp. sensor etc. that is black level drift upsetting the black point in the KCC files needed to grade the shot. If all that fails, you can always open the KCC files in DANCAD87.EXE (tm) and adjust the black level and curves in copies of the KCC files you need to use to compensate for the lack of matching "black frames", that is re-Grade for the data you have that is in one way of thinking of out-of-spec. Most of the time, if you use the commands as they were meant to be used, and you blow the dust off your sensor BEFORE you shoot, these steps should not require manual intervention other than running the commands that do the setup of the reference frames for the "dark field" and "bad pixel" removal. See Process Methods 20006, 20007, and 20008 for manual processing of the "dark field" frame sets, those Process Methods are in the Engineering process methods sub-menu. --- KCC PROMPT REDIRECT OPTIONS There are several ways to locate the KCC files (sometimes also using file extensions WBC, EDR, and SCP). The most direct way is to use the prompts in the execute menu in DANCINEC.EXE (tm) to enter the name of a *.KCC in any folder or directory on any disk by using its filename. The filename entered that way may also get filtered through the filename string substitution in the utilities menu (#5 from the main menu in DANCINEC.EXE (tm)) if you are using that to re-direct filenames to different drive letters etc. At those same prompts for the WBC KCC, the EDR KCC, and the SCP KCC filenames you can enter some reference strings to have the program seek the KCC files in the project shot folders. If you keep your WBC KCC in I11 level all the time, and the EDR KCC in I12 all the time, and the SCP KCC in I10 level all the time, then when you process a range of shots, the program can find the needed KCC by looking in those Ixx level folders in the shot folder of the project being processed. The project and shot values are taken from the source filename entered using option #1 in the execute sub-menu option #3 from the DANCINEC.EXE (tm) main menu. What you are entering is a type of "short hand" code string that starts with the @ symbol followed by some letters and numbers, and ending with .KCC extension. DANCAD87.EXE (tm)'s Grading command always saves the KCC files with the .KCC extension, but since Divided numbered filename type is (should be) used the *.KCC files will be in a sub-folder like KCC0000C\000.KCC and such in the Ixx image level folder of the Sxxxx shot folder. The point being that if you want to copy a *.WBC, *.EDR, *.SCP file and edit it you should rename it *.KCC, and if you are going the other way you need to rename it as well in reverse (not normally something users do, as the presets are calibrated when I hard code them into the menus for the presets for each camera which includes many values in the program being overwritten including the names of the *.KCC files to use for the color corrections). When DANCINEC.EXE (tm) sees a KCC filename starting with the @ symbol, it de- codes the instructions for where to look for the "real" KCC file needed. The numbers above the examples are for you to be able to count the char position better and are not entered anywhere. The letters and numbers must be in the right char position for the program to de-code where you are asking it to look for the KCC file, if there is a mistake you will probably get an error message or the KCC will be no being there will ignored and the end results will not be processed as you want them to be. Examples: 1234567890 @I09D0.KCC Uses KCC of type "D" for divided in I09 image level folder frame number 0. See code letter O below for frame number seek mode. 1234567890 @I10PS.KCC Uses KCC of type "P for padded in I10 image level then seeks first KCC for keyframe (S for Seek). See code letter S below for frame number seek mode. 1234567890123456 @I09DN123456.KCC Uses KCC of type "D" for divided in I09 image level then looks for frame number 123456 (frame KCC0123C\456.KCC in folder I09 in the shot folder). See code letter N below for frame number seek mode. 12345678901234 @I11DLR000.KCC Uses KCC of type "D" for divided in I11 image level then looks for Keyframe number of I11 folder in Edit List (L code for List) using the Edit list in the Reel 000 folder of the project folder (for the result file being made). Each Reel folder in the project can have its own version of the Edit List, so you need to tell the program which Reel folder to look for the Edit list in. This way of finding the KCC filename interacts with DANCAD87.EXE (tm) since when you change the keyframe (with the Edit list Pick command) and then use the Grading command to make a KCC file for the result in the result folder Ixx level, this option will "look up" the right keyframe number from the Edit list, where as the other options are not in "sync" with the value used in the Edit List so you need to manage them yourself. See code letter L below for frame number seek mode. 12345678901234 @I11DER000.KCC Uses KCC of type "D" for divided in I011 image level then looks for Keyframe number in the Edit List Keyframe frame number for the shot being processed in Reel 000 folder, then adds the frame offset. See code letter E below for frame number seek mode. If you don't want ANY *.KCC file used for the WBC, EDR, or SCP from the execute menu option #3 KCC prompts enter "DEFAULT.KCC" without the quotation marks and the program will not use that color correction, unless maybe you are using another KCC redirect or seek override, or you are using RAW data with meta-data with the color matching enabled, in which case the program would override the option #3 KCC filenames anyway. (Some camera models meta-data can override the *.KCC name redirect when you have that meta-data override enabled through the camera model selection sub-menus.) The @ redirect strings have these letter command options. For numbered filename types: L = Long numbered filename like: 0.KCC or 12345.KCC (Long is only used when other programs need it). S = Short numbered filename like: KCC.0 or KCC.123 (short is not used normally so don't select it). D = Divided numbered filename like: KCC0000C\000.KCC or KCC0005\989.KCC (Divided is used most of the time in result folders). P = Padded numbered filename like: 00000000.KCC or 00065534.KCC (Padded is sometimes used for results where other programs are involved, and Padded is frequently used for original source files since camera makers software does not normally make my Divided numbered filename type, but can sometimes make my Padded (or Long) numbered filename type). For frame number seek mode: O = frame number for KCC file is always 0 (frame number zero) (in the format of the numbered file type). O stands for zero here, always the same One number. This is the letter O not the number 0, like in One. S = seeks first matching frame number for KCC, if you change the keyframe you should delete any older KCC manually to avoid having the program find the wrong KCC file this way. S stands for Seek frame number here. This takes some time since it tries all the filenames numbered from zero upwards to a maximum value that would be longer than typical shots. N = looks for a the specific frame number entered as the number after N in the @ string, that is a literal frame number. N stands for Number, that is the number included in the @ redirect string but adjusted for the shot folder and Ixx level given. The shot folder changes depending on the source filename entered and if you enter a single shot or a range of shots with the source filename prompt i.e. #1 in the Execute menu. K = reads the keyframe number from a file in the shot's folder, the filename with the keyframe number is always "KEYFRAME.TXT" without the quotation marks, that way ALL shot folders can have a KEYFRAME.TXT filename file in them. This does not seek a KCC file, nor does it read the edit list file, it just looks up the value in the file names KEYFRAME.TXT, so you might edit that value in the(those) KEYFRAME.TXT file(s) manually if you need to. K stands for Keyframe named file. L = reads the keyframe number from the current Edit list value for the keyframe in the shot folder. The Edit list is found in the stated Reel level using the R command char in the @ string. The Keyframe is the same for all Ixx levels, but the KCC is saved by the Grading command to the result folder when you are grading, so the stated Ixx level tells the program which Ixx level to look in, as there might be 12 (twelve) KCC files with the same keyframe number in each of the 12 Ixx levels for each shot in the project, so the I command letter is needed along with the R for Reel when used with L for List seeking of the right keyframe number in the Edit list data. See the @I11DLR000.KCC example above, in that example I is for the Ixx image resolution level folder, D is for the Divided file numbering type, L is for seek key frame in edit List, and R is the Reel folder to find the Edit list in to open and check the data for the shot that is going to be processed. A = Like O but plus frame offset value. B = Like S but plus the frame offset value, that is it adds the frame offset to the number of the KCC found, so if you have the WBC, EDR, and SCP KCC in the same Ixx level on successive frame numbers you can use +1 and +2 to find the second and third files since seek will only find the first file in that case. C = Like N but plus the frame offset value. D = Like K but plus the frame offset value. E = Like L but plus the frame offset value. The frame offset value comes from the internal code of the process method execute command. If the @ redirect string is in the WBC prompt (#3 in the execute menu), then the frame offset is 0 (zero), i.e. no offset, the Keyframe value etc. If the @ redirect string is in the EDR prompt (#3 in the execute menu), then the frame offset is +1 (one), the first frame after the Keyframe value etc. If the @ redirect string is in the SCP prompt (#3 in the execute menu), then the frame offset is +2 (two), the second frame after the Keyframe value etc. The DPX KCC files for input and output should normally be in the DANCINEC.EXE (tm) program folder, as should the color matching ones used with the "presets" and the presets used by the RAW meta-data header (LK5) color matching options (matching between the cameras EVF "look" and the end BMP/CIN/DPX/TIF files). When the KCC prompts are used with just KCC filenames, and you process a range of shots all the frames get processed with the same KCC files, which might work if all the shots were from the same setup and no exposure changes or lighting changes were made etc. but if you need shot to shot processing then using that method you would need to setup a GVN file for each shot and make a BAT file and execute that so that the WBC, EDR, and SCP KCC files would change shot to shot using the individual GVN files names appended into the BAT file (the BAT file making command is in the execute menu). Although the BAT file method can be useful, it would be simpler for the program to just find the right set of KCC files for each shot folder and process the frames within the range as needed, such as doing a re-grade on all the shots in a project after editing. These @ redirect string commands can be used to get the program to automatically find the right KCC files when you process a range of shots (shot folder range is selected by the source filename prompt in the execute menu of DANCINEC.EXE (tm)). If something does not seem to be working right you might try contact me by email about what you are trying to do. Another option for having DANCINEC.EXE (tm) try to find the needed *.KCC files for you shot-to-shot color corrections is to use the Edit List Filename sync options listed in the next section. --- EDIT LIST FILENAME SYNC OPTIONS Although the re-direct options for the KCC files above used in the KCC filename prompts can be useful, some additional "sync" issues come up between DANCAD87.EXE (tm) and DANCINEC.EXE (tm). DANCAD87.EXE (tm) in its current state is not very fast in making the needed PIX files for sync playback of the shots in the NLE (Non Linear Editing) of DANCAD87.EXE (tm). Since DANCINEC.EXE (tm) is much faster in some ways in making PIX files (you can piggy back making PIX on BMP\CIN\DPX\TIF by adding 1000 to the result file type code, like 1036 for 16bpc DPX plus PIX, or 1048 form TIF plus PIX, made using the same RAW processing to cut total processing time in about half or so). That is all well and good, but in DANCAD87.EXE (tm) the edit list gets updated when new PIX files are added to the project folder's shot folders. To also have DANCINEC.EXE (tm) update the Edit list to show PIX files added, and so let you start editing those shots without having to manually enter the shot length DANCINEC.EXE (tm) needs to be told a few things first: 1) What Drive the project folder is on to update the edit list in. (like C:\) 2) What the Project number is. (Each movie is the same "Project" for all files no matter what disk the Pxxxx folder is on, as you can have ancillary Project folders on many disks to increase the TB storage, in the Edit List you can "map" the drive letters as needed to tell the Edit List what disk drive to look for another Project folder on, you can scroll the "spread sheet" like Edit list off the left of the screen to see the prompts for the twelve Ixx levels, and likewise in the audio Track lists.) 3) What the Reel number is that the Edit list to update is in. (Reel R000 is the whole film for editing, later R001 and up can be used to cut the Reels for filmout to 2000' rolls of film by making a copy of the Edit list in Reel R000 folder). 4) The result shot folder (can be taken from the result name prompt). 5) Is the shot folder (or shot folder range) a reference to the physical shot folder number, or the edit shot number. Most of the time DANCINEC.EXE (tm) is used to process RAW data on physical shot folder numbers because you would load shots into consecutive folders before re-arranging the shot numbers in editing. But because you might want to grade shots AFTER editing, the logical edit order of the shots in a range would be of more use. You can select both ways in the Edit list sync options menu. For logical (edit) shot folder order to work, DANCINEC.EXE (tm) needs to open the edit list and look up what physical shot folder holds the logical shot's files. Once you have told DANCINEC.EXE (tm) how to find the right Edit list file for the project you want to process image files for, you can set the "sync" options active or disabled in the "Edit list sync" sub-menu #6 in the Execute menu (#3 from the DANCINEC.EXE (tm) main menu). The options in the "Edit list sync" sub-menu are: 1) Reset sync options to disable. This turns off ALL the sync options so that the KCC prompts and other "normal" setup functions work without any options in the "Edit list sync" sub-menu overriding them. 2) Select Drive, Project, and Reel to sync with. This is needed since DANCINEC.EXE (tm) can operate on and with source and result files outside of any Project file Structure, or independent of the shot information in the Edit list for any given project, so you need to tell it what project's folder to look in to find source files and to put result files in and what KEL (Kinema Edit List) file to alter or read shot information from (start frame, keyframe, last frame). 3) Sync Test option to Key-frame for shot in Edit list. Normally DANCINEC.EXE (tm) defaults to the execute Test option to process frame 0 (zero) in a numbered set of frames. If the first (0, zero) frame is a flash frame, sync mark, or otherwise not usable as a reference frame for Grading, the command #7 in the Execute menu lets you change the Test frame number to some other frame in the shot. Say you have a shot of a blank wall and half way through the actor walks by, then if you know the shot is about 360 frames long, then you would want to use a frame near 180 to look at to judge the grading. DANCAD87.EXE (tm) was setup to take care of that issue automatically by selecting a "Keyframe" between the head and tail trim points for each shot, that "Keyframe" number is stored for each shot in the Edit list. When you use this sync option the Test frame number entered with #7 in the Execute menu is ignored, and the Keyframe number is looked up in the selected Project and Reel Edit list. That way each shot can be worked on and you can look at Test frames of the Keyframe you selected for each shot when editing, its all automatic you don't need to remember the right Keyframe or enter its value automatically each time. If you did not pick a Keyframe the Edit List defaults the Keyframe to the Head frame made when you trim the shot length, if you did not trim the shot yet, then the default is the zero frame, so you can disable this sync option, then the number you enter in command #7 in the Execute menu selects the frame number to look at, later when you have time you can use DANCAD87.EXE (tm) to mark a Keyframe in each shot of your project so that you can have this sync option active and see the marked Keyframes each time you use the execute Test with full screen Graphics "preview". In the first release the "preview" does not work in the GUI version, so use the Windows Console or DOS versions. If you can only use the GUI version, you can still test, but need to use another viewer to see the result file. 4) Sync KCC to Key-frame selected in Edit list. When you use the Grading command in DANCAD87.EXE (tm) the KCC file saved is normally for the frame being graded from, which would be the selected Key-frame, and the KCC made by the Grading command goes into the result folder Ixx level, so when you are done grading a shot the KCC file(s) are stored with the Keyframe number in the result folder(s). This sync option lets DANCINEC.EXE (tm) look in the Edit list and figure out what folder the KCC file needed is and under what frame number it was stored. This then overrides the KCC prompts in #3 in the Execute menu to have each shot in a range of shots Graded with the right KCC files for just that shot in the range of shots (without sync to the Edit list, all shots in a range being processed would be Graded using the same KCC file from the #3 prompt in the Execute menu). You should keep in mind that if you change the Key-frame after grading the edit list will no longer point to the KCC files, and so if you process frames at that point the results will not look right. If you do need to change the Keyframe because you moved the head and tail edit points (Keyframes must be within the edited frame range because when you use Pick to view the frames playback the keyframe must be within the playback range you trimmed the shot to) you must use the Grading command to "Clone" the previous graded frame to the new Keyframe, or manually renumber the KCC files for that shot. Normally you would not do the final grading until after you have trimmed a shot and selected a Keyframe from within that trimmed frame range, and would not change the selected Keyframe as it is your reference for that shot. To avoid issues with the Keyframe number changing after you have Graded, try to select a Keyframe in between the head and tail enough that if you need to adjust the trip points you do not upset the Keyframe between them. 5) Sync edit list to range of PIX frames made in DANCINEC.EXE (tm). Both DANCINEC.EXE (tm) and DANCAD87.EXE (tm) can make PIX playback frames for use in editing with DANCAD87.EXE (tm). But DANCINEC.EXE (tm) can make them much faster from some kinds of source files and for some video modes. If you are making PIX frames for editing from your camera's RAW data you would use DANCINEC.EXE (tm) because it is much faster than doing a de-Bayer to TIF or BMP and then using DANCAD87.EXE (tm) to convert those TIF or BMP to PIX. But there is an issue, when you make the PIX in DANCAD87.EXE (tm) it updates the Edit list so that the range of frames from head frame (zero) to tail frame (whatever the source shot frame count was) is entered for you automatically and you can then go directly into the Pick command from the Edit list to start editing. This sync option is to add that function to DANCINEC.EXE (tm), that is when it makes PIX frames it updates the Edit list so that when you enter the Edit list to do some editing on the "new" shots added, the frame numbers for the head (0) and tail (number of PIX made) has been updated for you automatically for all the shots in the range that you added to the Project. You can "piggy-back" the making of PIX frames for editing on making BMP/CIN/DPX/TIF frames in DANCINEC.EXE (tm) to save time, just add 1000 to the result filetype code, like 1008 for monochrome BMP + PIX, 1024 for color BMP + PIX, 1029 for CIN + PIX, 1030 for 10bpc DPX + PIX, 1036 for 16bpc DPX + PIX, and 1048 for TIF + PIX. The setup for the PIX frames to make is in the global setup menu #1 from the DANCINEC.EXE (tm) and in DANCAD87.EXE (tm) main menu when you press [~] key. The setup for the PIX frames depends on the speed of the computer you will use DANCAD87.EXE (tm) with and depends on the video resolution used, 640x480x16bit color mode being the "best" choice for most computers to avoid dropped frames on playback, the PIX are letter-boxed or stamp-boxed and you pick size within the screen size limits that does not over-tax your computer's drives speed to keep up during playback. Higher resolutions and color depths are supported up to 1920x1440x32bit, but I don't know of any computer that can display uncompressed frames that size at 24fps, 25fps or 30fps. The highest you might get working would be 1280x720x16bit (from RAID SSD). For Black and White movies the 8bit monochrome mode would look better than the 16bit color mode (which needs to be dithered so looks grainy) and they are half as big should should play better. (DANCAD87.EXE (tm) uses VESA VBE 1.2 banked memory mode, not linear mode as DANCINEC.EXE (tm), DANCINED.EXE (tm) and DANCINEL.EXE (tm) can, so although in theory it supports 2048x1536x32bit in its code, I have not found a video board that supports banked mode that works right yet, the video card BIOS don't seem to be bug free for that resolution in banked mode. Even if you found a graphics card that worked, the computer would probably not be fast enough for 24fps playback at the full 2048 width. Some GeForce (tm) cards may display part of the screen in 2048x1536 banked resolution.) 6) Sync range of frames to process in DANCINEC.EXE (tm) to the Edit list range of frames for the shot being processed. This sync option would be AFTER the shots have been trimmed in DANCAD87.EXE (tm) and no further changes would be made to the editing. Its to save time by only processing the frames needed based on the edited range of frames in each shot. If you need to expand the range of frames in a shot later, then you would need to re-grade the shot because frames outside the range that was in the Edit list the first time you graded only the edited frames was smaller so the Grading of the added frames would not match. The original vision for how things were to go was to Grade ALL the frames for all the shots needed so that edit changes could be done at any time and all the frames needed would always be ready to be used in the final "cut" of the project. The problem with that is that unless you "pre- trim" the frames in the shots as you insert them into the project shot folders, you would end up with more than double the frames needed later for the edited down version. Because working with 4K frames takes a long time to process all of them, and several computers are needed, the savings from ONLY doing full finish processing on the frames needed in the final "cut" can more than outweigh the issues that come up, you could save quite a bit of money and months of time this way. So I added this option to make the programs more productive, its just a but of more planing out the workflow steps when using it in order to save time and money later. You can use this sync option or not as you please, it is not required for the workflow to function, but could cut the finish processing time to less than half which is no small issue since at higher image resolutions the finish time grows by the square of the image K value, if you do it at 2K then do the same project at 4K it takes 4x as long, and at 8K 16x as long and at 16K 64x as long and so on. That is one reason that working on movies at 1920x1080 is common, and 4k less common, and 8k almost not at all (IMAX (tm) seems to scan at 8K then DI at 4K from what I have read from unofficial sources). 7) Select using "shot folder order" or "edit shot order" for shot numbers. When not using the Edit list sync options DANCINEC.EXE (tm) can ONLY use the physical shot folders literal names, S0001 for shot folder 1 in that project. Since in editing you need to be able to move shots around, in DANCAD87.EXE (tm) the shots have "two" shot numbers each, the physical shot folder number, and the logical shot number in the project after editing and moving the order of the shots around. So after editing you can get something like this: Shot 1 in edit order = Shot 785 in folder order Shot 2 in edit order = Shot 4 in folder order Shot 3 in edit order = Shot 89 in folder order Shot 4 in edit order = shot 5586 in folder order When you first load several shots from a says shooting you could load them into the project in consecutive physical shot folders, so you say maybe todays shooting goes into physical shot folders 500 through 700. You can then do a "one light" grade on all of those shots 500 through 700 to produce PIX files for editing. Later after you have done your editing and graded on the Keyframes selected for each shot, you would want to Grade the range of frames in an edited sequence, which would not be in consecutive physical shot folders. This Edit List sync option is to take care of that issue, when you are grading "dailies" you can pick a range of physical shot folders to process, but when you are doing final Grading one edited sequence at a time, you can select "edit shot order" and have DANCINEC.EXE (tm) look in the shot list to figure out how to translate edit shot order range start shot to end shot, into physical shot folders of any number as "moved" around during editing. In BOTH cases, the source filename prompt #1 in DANCINEC.EXE (tm)'s Execute menu lets you set the shot range, this sync option determines how THAT range is interpreted during process method execution. Be sure to reset this to the right mode to avoid overwriting frame files accidentally, as it could take a long time to re- generate them if you have this mode set wrong, and if you don't have backups you might lose important files if non-backed up data gets overwritten because you had the mode set wrong. When working with your frames you should make backup copies of everything, later when you have finished your project, you can decide what needs to be kept for archival use and what you can erase so the disks etc. can be used over again or sold to recoup costs. Obviously the Edit list options would not be of interest to people not using DANCAD87.EXE (tm) since that is the program that makes the Edit list, DANCINEC.EXE (tm) just updates and reads from the existing Edit list KEL files. DANCINEC.EXE (tm) is designed to be able to be used by itself for frame processing without needing to use DANCAD87.EXE (tm) for your editing. --- CAMERA PRESETS FOR COLOR CORRECTION Presets for the following cameras based on what sample files I had on hand were made: 301xxxx.* Kinor-2K (tm) some DNG from a Russian Digital Cinema Camera that makes DNG frames. 401xxxx.* Acam dII (tm) some DNG from a Swedish Digital Cinema Camera that shoots RAW and makes DNG files. 601xxxx.* SI-2K (tm) a DNG from the Silicon Imaging (tm) Digital Cinema Camera that shoots compressed Cineform (tm) and/or RAW and can make DNG files. 701xxxx.* A sample frame from a camera under development that shoots True RAW sensor data saved as RAW data (without header). Other preset samples may have been added since this document was written, check DANCINEC.EXE (tm)'s menus to see what is in the current revision. The sample images I had to work with were not shot under well controlled conditions and so may not give best results, if you have some sample frames of your shots you might contact me about doing a sample correction so you can see better what your results might be with the adjustments closer to what they should be for your exposure level and camera type. I was unable on program release (about April 2011) to include a preset calibration for the SUMIX (tm) camera because the RAW sample file I have was mucked up with poor white balance, noise reduction, and gamma correction in the SUMIX (tm) camera software. To make a proper preset the data would need to be shot without those processing steps active so that actual sensor data would be what is recorded. The presets are in the process method menu option #6 in the process method menu #2 from the main menu in DANCINEC.EXE (tm). Using presets can "free" you from having to use DANCAD87.EXE (tm)'s Grading command to make manual adjustments for each shot's key-frame, but it can only give you a short of "one light" grade and requires that you are very careful about the original exposure of your RAW data to reduce color correction mismatches. You can re-grade the RGB result files, but its better to get as close as you can in the original color correction done during de-Bayer as the various filters used during the de-Bayer work best when the color correction is close to the end use values. I did not have "calibrated" reference frames for the camera's named above, not that I did not ask for some from some of the camera makers, in order to make exact 5500K and 3200K presets, rather I tried to get a balanced result based on their sample files, which were shot at some unknown EI/ISO and lighting type. If you are trying to use DANCINEC.EXE (tm) it would be good for you to email me about sending me some calibrated reference frames shot under known K value lighting and at some known EI/ISO speed so that I can look into adding presets of more general use to your style of shooting and that others might find useful as well for their shooting with the model of camera you are using. --- META_DATA "LOOK" COLOR MATCHING PRESETS Right now (about September 2011) only one camera company has worked with me to get the "round robin" for in camera "look" tables to function with my color correction so that there would be some match between the camera's viewfinder and the automatic selection of presets from meta-data on each RAW or DNG data frame. That company is Kinefinity.com (sm) and the DNG plus meta-data (my model code 504) relates to their KineRAW (tm) camera series with various size sensors. They have two data formats for meta-data, my camera model code 503 for RAW+meta_data and my camera model code 504 for their DNG plus meta-data contained in the TIFF/DNG Tag 50740 area. It seems the 503 file format is not going to be supported by Kinefinity.com (sm)'s production cameras, so when meta-data is talked about in my programs and documentation it is probably my camera model code 504 that is being spoken of, but both 503 and 504 meta-data are shared by some program operations as they are similar, with some extra functions added to the 504 meta-data. When processing the DNG files made by their cameras, and using other de-Bayer programs the meta-data in tag 50740 may just be ignored, so if you want it interpreted you may need to my use DANCINEC.EXE (tm) et al. It should also be possible to select those same presets manually. The presets are in the process method menu option #6 in the process method menu #2 from the main menu in DANCINEC.EXE (tm). See Kinefinity (tm) camera model code 504 (DNG plus meta-data) in the camera selection menu #3 in the setup menu #1 from DANCINEC.EXE (tm) main menu for the meta-data header enable/disable settings. This model camera's meta-data can override some process method functions and so you need to be mindful of what features you are overriding or not as you will become confused when prompts elsewhere in the program do not seem to be working because you forgot that you have set this camera's meta-data to override them. If you need to keep notes on paper as to what settings you have setup for the meta-data overrides, but most of the time once you setup the program to work the way you want it to you would be batch processing most of the frames that way and so would not need to make frequent changes to these meta-data override flags, maybe. Information for guide lines used in the camera's viewfinder can be passed through the meta-data to the "workprint" quality results (or finish) by way of the guide elements burn-in. Likewise you can burn-in SMPTE time code, and the Date+Time Of Day and other fields in the meta-data as text elements, or you can pass those to DPX or CIN files made so that the camera time code and Date+TOD are in the headers of the DPX or CIN files you generate for use in other editing and DI systems. Because of limitations in the real time processing of the camera's viewfinder hardware and EVF or LCD monitor etc. there may be differences between the in- camera color correction for the camera's viewfinder and the "finish" color correction of the result frames produced by DANCINEC.EXE (tm). The "finish" color correction can include area filters, the EDR and SCP KCC steps and so may show a bit more highlight detail, less noise, greater sharpness, and a more "film like" result, perhaps, so hopefully the differences are in a more positive than negative direction. For more information about the Kinefinity.com (sm) KineRAW (tm) cameras look over their web site and downloads if any. The camera's menus should have some options that relate to the meta-data saved into Tag 50740 in the DNG frames their cameras make. You may need to update the *.LK5 files for the viewfinder "looks" in order to get good color matching with the finished de-Bayered TIF/DPX/CIN/BMP frames you make with my programs. DANCINEC.EXE (tm) has commands that make the *.LK5 "look" LUT files for their cameras, see the process methods Engineering menu. You need to execute the reference frame's processing to make the *.LK5 file, which means that you need to shoot a reference DNG at the right EI/ISO and K values of light and exposure in order to color correct the WBC KCC file in DANCAD87.EXE (tm) so that the WBC KCC file can be used to make the LK5 file. The reference DNG frames should contain both a Kodak 90% white card as well as the Kodak 18% Gray card, along with a color checker chart like the MacBeth color checker chart so you can set the color saturation using the chroma matrix values in DANCAD87.EXE (tm)'s Grading command control screens while making the KCC that gets processed by DANCINEC.EXE (tm) into the LK5 file for the camera's viewfinder to use so that you see the right EI/ISO, K value as well as light type and "look group". --- DPX LUT, GAMMA AND CLIP POINTS To give fine control for import and export of data in DPX and CIN files, there are two KCC files, they let you adjust the clip points and curves. Those are selected with the execute menu KCC names option #3 in the execute menu in DANCINEC.EXE (tm). In addition to those there are programed hard clip and gamma adjustment commands in the setup menu #1 from DANCINEC.EXE (tm)'s main menu. They let you set the hard clip points, e.g. 95 for black and 685 for white, and the gamma conversion for the Cineon (tm), aka Log-C Log curve which is about gamma 1.68 to 1.72 so you can view the test results on a gamma 2.4 monitor etc. Standard encoding gamma for HD images is gamma 2.22 BUT the standard monitor gamma is gamma 2.4, so they are not the same, and DCP projectors can be gamma 2.6 which is said to be done because the movie theatre is darker than where computer monitors or HD TVs are viewed in higher ambient lighting. Gamma 2.6 shows mid-tone a bit darker than on a monitor at Gamma 2.4, but the image frames made for both uses would have mid-tone set for Gamma 2.22 or 0.462 of full signal. Older SD standard of gamma 2.2 for video signals would have mid- tone at about 0.458, but as things are all HD now, its better to grade mid-tone for the newer HD standard. Some LCD monitors, HD TVs, and other ways of looking at images may not be at the monitor standard of gamma 2.4 so may show the mid-tone too light or dark so its best to use the test probe in DANCINEC.EXE (tm) and DANCAD87.EXE (tm) to check the image data numeric values to see if you have mid-tone at the right value of 0.462 and such (0.462*255=118, in 10bpc DPX mid-tone is 470 (445 is the D-LAD tone that is lower than 18% gray card, in the film LAD they used a 16% gray for the LAD patch, not a 18% gray card it seems according to some notes in a Kodak pdf file.)). In the process method menu is a command that can "tag" a CIN or DPX file with gray steps and patches so you can try to use the brightness probe in the graphics Test screen in DANCINEC.EXE (tm). The Cineon (tm) type images can be a bit confusing unless you have some reference patches to check the effect of curves on, such as softclip settings and gamma adjustments, so the "tag" command can be a big help of knowing where mid-tone should be, that is if you shoot a test frame on your set and hold an 18% gray card in front of the camera, then process the RAW data and want to end up with a Cineon (tm) 10bpc DPX result, the gray card should be brightness code R=470 G=470 B=470, but the image tones are NOT the same values as you would see if you did not correct the display gamma in the display LUT, so to get the display LUT to read right, you "tag" the Kodak D-LAD CIN file, measure the patches that are added with the Test display probe, then process your shot's test frame and match the result levels, which with the display LUT set to default would be values other than normal encoding for gamma 2.22 images (if the display LUT is set right, then even though the output DPX is right for Cineon (tm) Log brightness, you will measure the right mid-tone for display on gamma 2.4 monitors). You need to remember if you are viewing through the monitor LUT or not when working with DPX files since you cannot get a "normal" looking display without using the monitor LUT, if you did then the file would not be "right" by the Cineon (tm) standard levels for the tones (that is the Cineon (tm) "correct" image should be gray and washed out looking if viewed without a monitor LUT as say a BMP file would be)). The Graphics display part of DANCINEC.EXE (tm) has its OWN setup for the monitor LUT and clip points and gamma for loading and viewing CIN/DPX files (as wells as for the ITU/CCIR 601 limits for BMP files) so you need to setup the Graphics *.FIG file for your display choice and ALSO setup the load and save tonal adjustments for CIN/DPX (as well as BMP) and keep in mind that you do not want to "un-Log-C" the images TWICE, once in the frame processing, and again in the Test view of that processed result, or you will see very messed up images that do not match the downstream programs when you use the Graphics Test to see your results. So there are actually TWO places you setup these LUT settings, in the #1 setup menu, and in the #4 graphics part of DANCINEC.EXE (tm), and both need to be done so that they work together to view what you want to see, that is if you are making DPX with "Log-C" encoding, but you want to see what it will look like in the projected movie, you would set the graphics to "de- Log-C" the image to monitor gamma 2.4, and such. Kodak (tm) seems to have picked Cineon (tm) values that do put 18% Gray Card near where it would be for gamma 2.2 to 2.22 adjusted images if you have black clip at 0 and white clip at 1023, 470/1023=0.459 and 18% at gamma 2.22 comes out to about 0.18^(1/2.22)=0.462 but values higher and lower will be too light in the shadow areas and to dark in the highlight areas without the 95 to 685 clipping. When you subtract the 95 from the DPX values you get: 685-95 = 590 470-95 = 375 375/590 = 0.6355 0.6355 is over the gamma 2.22 display value of 0.462, so some curve adjustment is needed to bring mid-tone down to the right value after clipping CIN/DPX files. That can be done with the CIN/DPX input KCC file, or with the CIN/DPX input gamma adjustment. 0.6355^1.7 = 0.4626 So you can use 1.7 at the CIN/DPX input gamma adjustment prompt to convert mid- tone for use on normal monitor gamma 2.22 displays. When you use the same values for the input and output clip points and gamma, the program is setup to do the 1/gamma for you automatically so you don't need to remember to figure the inverse values, that is you use the same value for the CIN/DPX gamma in the input and output configuration if you want to have the output file match the input file Log state. Remember to turn OFF the monitor LUT in the graphics test view setup (in #4 from DANCINEC.EXE (tm) main menu) if you want to measure the "Log" CIN/DPX values with the "probe" otherwise you will be looking at a file that was clip and gamma converted twice (since you normally have the graphics monitoring LUT set to convert Lot to Monitor Gamma 2.4 in order to look at Log CIN/DPX results and have them look like they would on the movie screen with full tone and contrast, rather than gray and washed out in their Log state). See also additional information on CIN/DPX files below in the Process Method 95510 that "tags" CIN/DPX files with gray step charts so you can better figure out what is going on during Grading and use the test Probe to measure the tone values you get with various curves and gamma adjustments. --- DPX HD LIMITS DANCINEC.EXE (tm) et al. can make "HD" type DPX files in 10bpc and 16bpc in addition to Cineon (tm) Log-C type. The clip points are like the ITU/CCIR 601 limits but adjusted for the bpc level so: 8 bpc, Black clip = 16, Midtone = 118, White clip = 235 10 bpc, Black clip = 64, Midtone = 470, White clip = 940 16bpc, Black clip = 4096, Midtone = 30080, White clip = 60160 In general the 10bpc levels are 4x the 8bpc levels, and the 16bpc levels 64x the 10bpc levels. The values for mid-tone are approximate you can get them like this: 255 * 0.462 = 118 1023 * 0.462 = 473 65472 * 0.462 = 30248 or (235 - 16) * 0.462 = 101 + 16 = 117 (940 - 64) * 0.462 = 405 + 64 = 469 (60160 - 4096) * 0.462 = 25901 + 4096 = 29998 Level for 90% white card goes between the midtone and white clip levels, but its level is not defined and depends on the exposure of the camera sensor or film stock and how much highlight detail above 90% white you want to keep under white clip level, in other words there is no upper limit to brightness and at some level above 90% white you need to clip yourself or take the clipping of the sensor or scanner. Film stocks can record very high brightness highlights on their S-curve when processed right, and so have no fixed upper limit unless they solarize then it would be the peak density point on their exposure vs. density curve. For HD files it is assumed that midtone is an 18% Gray subject and that it is adjusted to come out at about 0.462 of full signal. 0.18^0.45 = 0.462 0.45 = (1/2.22) 2.22 is the gamma correction standard for television images, and some computer images such as for use on the internet. The natural gamma of CRT monitors was about 2.4, and some DCP projectors are about gamma 2.6. But to compensate for the ambient room or theatre light level at which TV, computer, and digital projectors are used at the images are all encoded for gamma 2.22 correction, which is gamma 0.45. Gamma 0.45 means that 0.18 signal level is encoded as 0.462 signal level, or the mid-tones are lighter than in a linear image as gamma is a curve that meets the white and black levels, less change in brightness takes place in the highlights and more in the shadows. To make HD output files that are range limited, you need to adjust the KCC files used, such as the DPX output KCC file, so that the three points align with the needed levels shown above for black, midtone, and white. Between black and midtone the curve should be close to the gamma 0.45 curve, but between midtone and white some "extra" bending down of the highlights may be needed if you are starting with "linear" sensor data that was under-exposed because you will have quite a bit of more of the sensor range above 90% white so you would put the 90% white card BETWEEN the midtone and white clip levels shown above, NOT AT the white clip level. The actual signal level for a 18% gray card would not be 0.18 signal since the sensor has a black level of about 0.04 signal level, and the exposure of the sensor is set for Digital Cinema Camera use such that the green pixels exposure for a 90% white card is at between about 0.04+0.25 signal level and 0.04+0.50 signal level, and for high contrast subjects the 90% white exposure may be as low as 0.16 above black bias or 0.04+0.16 = 0.2 absolute sensor signal level. The 0.18 signal level would be for after the sensor black bias is clipped off and the range normalized, and at "video" exposure levels that are higher than used for Digital Cinema Cameras being the highest exposure level possible to minimize sensor noise in broadcast video cameras, well at least in the past when sensor noise was an issue, today most professional cameras expose the sensor well below its maximum so that white parts of the subject are not burned out by sensor, pre-amp, ASC clip, and so the actual sensor levels you see in your RAW data will need gamma 0.45 lift as well as additional S-curves applied to bring mid-tone up to the needed levels without pushing the highlight detail above 90% white card off the top of the range, that type of S-curve is called "soft clip" or "hyper-gamma (tm)" or other trade names. With my software you can make any type of S-curve or such you want since the Grading commands in my CAD programs let you adjust the curves used for making the color corrections and EI/ISO mid-tone compensations. --- BMP ITU/CCIR 601 LIMITS Provision has been made for LUT to convert the image data both on import and export from full range 0 to 255 8 bpc data to the ITU/CCIR 601 limits of 16 for black and 235 for white. Those are in the setup menu #1 from DANCINEC.EXE (tm)'s menu. When the limits are active on export the image is lower contrast and lacks any true white or black, but that is the standard for some video display options so some programs downstream in the workflow expect that, if you do not "flatten" the images before export your DVD or other use may show blown out highlights and a loss of shadow detail when viewed on a TV set etc. You need to know if the downstream program will itself apply the 16-255 tonal value range since you do not want to output result BMP that are reduced from 0- 255 range down to 16-255 range if the downstream program is going to further reduce the tonal range which would result in the final viewing of your images not having ANY monitor white or black shown. --- USING OTHER RAW CAMERAS SUCH AS HDSLR Adobe (tm) makes a free RAW to DNG converter program, Adobe Digital Negative Converter (tm), to convert so called "RAW" camera formats from various camera makers such as "Canon (tm)" or "Nikon (tm)". If you want to use DANCINEC.EXE (tm) to convert say *.CR2 image single frame files shot in a Canon XTi (tm) used for stop motion, go-motion, claymation, time lapse, or in a film scanner etc. you can use the Adobe (tm) converter program to first convert the *.CR2 files into DNG, then use DANCINEC.EXE (tm) to convert the RAW data part of the DNG file into BMP/CIN/DPX/TIF RGB files with full processing and color correction (or workprinting quality as you like.) When you run the Adobe (tm) RAW to DNG converter program be sure to find the prompt that selects Bayer DNG or Linear DNG result, and select Bayer DNG, in the version I tried it was in a second level menu: 1) Click on Change Preferences Button. 2) Uncheck Compression if its is checked. 3) Click the "radio button" marked "Preserve Raw Image" and make sure the "radio button" dot is not selected for "Convert to Linear Image". "Preserve Raw Image" is for moving the Bayer data from the RAW file type for your camera to DNG without changes. That is not the prompt for embedding the original file in the DNG, as that is not something that is wanted or required for use with my programs. You do not want Linear DNG, only Bayer DNG holding the sensor data uninterpolated. Linear DNG is interpolated and should not be used as having the converter interpolate the data could de-grade the result. If you cannot make a Bayer DNG, then use the free Adobe (tm) program DNG_validate.exe (tm) to convert the Linear DNG to a TIF file, then process that TIF file using the RGB to RGB process method in DANCINEC.EXE (tm). Select the generic DNG support in the camera menu to read such Bayer DNG files. If the program reports some errors contact me by email about the issue, I may then request a sample DNG file to see what the error might be. There is also a DNG header trace command in DANCINEC.EXE (tm) that might show where the issue could be. DNG files are not a fixed header format, and new "Tags" are added or used in new ways, so issues can evolve from that ambiguity of how the many TIF/DNG "Tags" are used in any given TIF or DNG file. As was mentioned, the Adobe (tm) program DNG_validate.exe (tm) can be used maybe as a last resort to get your images into an non-white-balanced 48bpp TIF file. The quality may not be as good going that path, but at least if no white balance is done, you can do your own white balance and tone adjustments in DANCINEC.EXE (tm) so the result may be better than using some other program to make the TIF files. You can also use the software from your camera's maker to convert the camera's RAW file type into 48bpp TIF, then use DANCINEC.EXE (tm) for additional processing or to make "opticals" etc. If you have Bayer DNG or 48bpp TIF files that will not load, you can contact me about the issue, if I have some sample files I might be able to see what the issue is. I do not support TIF files other than 48bpp 16bpc RGB at this time. --- USING COMPRESSED CAMERAS LIKE H.264 or REDCODE (tm) R3D (tm) DANCINEC.EXE (tm) only de-Bayers "True RAW" Digital Cinema Camera footage (or at least whatever can end up in Bayer format), if you are shooting movies with compressed cameras that use various "lossy" compression systems etc. you need to use the camera makers de-Bayer program or some other program using their SDK etc. and make BMP/CIN/DPX/TIF image frames. The best choice to make is 16bpc DPX or 16bpc TIF frames. If you need SMPTE time code for burn in etc., use 16bpc or 10bpc DPX as your conversion file type. Once you have made the needed RGB files, you can use DANCINEC.EXE (tm) to do color correction or "opticals" as needed by using the RGB to RGB process methods (not the de-Bayer process methods). If you can make an uncompressed AVI file from your H.264 or other compressed video format that your camera records using, you may be able to use VirtualDub (tm) to break that uncompressed AVI file into BMP frames. Since BMP frames are only 8bpc (24bpp) any color correction done on them will result in histogram gaps, but if you do "opticals" only without color correction, then the result should not develop histogram gaps as much. If you can try to use higher bit depth RGB file formats for intermediate steps. If your compressed file format is only 8bpc, then your grading range on that footage before histogram gaps become visible would be limited, not by my programs, but rather by the camera's original compressed file format not having enough tones to start with to allow much re-Grading without scaling errors above the 8th or 9th bit showing up as histogram gaps. When using camera's data that will pass through 8bpc file formats, try to get the exposure, color correction, look setting and lighting right at the time of shooting before the compression in the camera to minimize the need for additional color corrections later which would add histogram gaps when working with 8bpc images. There is a new "freeware" utility it seems called "5DtoRGB (tm)" that can do a high quality (?) conversion from the H.264 files made by some HDSLR to DPX 10bpc files. Since DANCINEC.EXE (tm) reads 10bpc DPX and can write 48bpp 16bpc TIF files, you should be able to convert your H.264 footage shot in a HDSLR into color corrected TIF files for use in DANCAD87.EXE (tm)'s CC/NLE/MIX and edit and build the sound track mix, then output edited frames for film transfer, DCP projection, and Blu-Ray or DVD production. Search also the third party software easyDCP Creator (tm), easyDCP Creator+ (tm), and easyDCP Player (tm). For REDCODE (tm) R3D (tm) type files, you can use REDCINE-X (tm) to de-Bayer them and make 16bpc DPX or 16bpc TIF files for processing in DANCINEC.EXE (tm). --- GRAPHICS TEST SCREEN PROBE KEY COMMANDS IN DANCINEC.EXE (tm) In DANCINED.EXE (tm) and DANCINEC.EXE (tm) if you can get the graphics working on your computer, there is a graphics test probe to measure the digital "density" of the color corrected images. When movie film was in use, optical densitometers were used to check that the film was processed and printed to the right density. Likewise the Probe in DANCINEC.EXE (tm) can be used to check the mid-tone and other digital "density" of your processed frames to check that you have things where they should be. The use of the probe is especially useful when working with DPX files since they are not for viewing at image encoding gamma 2.22 and monitor gamma 2.4 in many cases because of their Cineon (tm) aka Log-C about gamma 1.7 correction and their reduced black and white clip points. See the math worked out above. To help with processing CIN and DPX files, a special process method 95510 (in the RGB process methods special process sub-menu) is used to overlay square patches of various tones over a frame of your shot you are working with so that when you process that "tagged" frame with the color correction process methods you can then measure how the tones changed. Process method 95510 overwrite pixels in the actual CIN or DPX file, so the numbers for the tone patches are absolute, at least in the 10bpc files. In the 16bpc files the tones are scaled from the standard 10bpc file standards, that is code 470 is 18% gray card, code 95 is black clip, and 685 is white clip in the 10bpc files 0 to 1023 range of tones, in the 16bpc files its increased in ratio to the larger range. Kodak (tm) has its D-LAD patch set to code 445 for being density 1.03 about in the release print, so that large gray patch is NOT meant to match an 18% Kodak Gray card, why that is does not seem to be well documented by Kodak (tm) (?). Kodak (tm) does say that in the older film LAD, the center gray patch on the left of the frame was 16% Gray Card, so its also darker than 18% Kodak (tm) Gray card. Where 18% gray comes from is that if you have a monitor that is at gamma 2.5 then 18% gray should be half the display signal range, but since the newer HD TV and now the computer standard is gamma 2.4 for monitors the display signal should be about 0.462 of full range or about code 116 to 118 for 0 to 255 8 bit displays, or also for 8 bit displays reduced to the ITU/CCIR 601 limits of 16 to 235. Code 118 is close to 18% Gray Kodak (tm) cards for images encoded gamma 2.22 for viewing on gamma 2.4 monitors. So when you convert a CIN or DPX file that has a 18% gray card in it, two things should happen: 1) The 18% Gray card in the images should match the 470 patch in the "tagged" CIN or DPX file no matter what color corrections are made. 2) If you have the black and white clip points set right and the additional gamma correction, then when you measure the 470 patch and the image of the 18% Gray card off the screen with the probe in the graphics preview in DANCINEC.EXE (tm) you should read about 0.462 or 118 (in 0-255 8 bit range) with some small rounding errors because the probe reads 8 bit values now (there are various ways to display the probe reading for 8, 10, 12, 14, 16 bit range and in real number normalized range 0.0 to 1.0). The probe reading is taken from the screen pixels, and not the file made, because the read LUTs need to be in the signal path to compensate for the CIN or DPX being Cineon (tm) aka Log-C or Red709 or Linear Gamma 1.0 etc., or not as you like, but to measure mid-tone at encode target value of gamma 2.22 for gamma 2.4 monitor display corrected, the input LUTs for the CIN/DPX load would need to be active most of the time (see the graphics setup in DANCINEC.EXE (tm) via option #4 from its main menu to turn on or off the input LUT for CIN/DPX on viewing, for the LUTs processing the actual data IN the CIN/DPX see the sub-menu in the setup option #1 from the DANCINEC.EXE (tm) main menu.) When you get a CIN/DPX file to look right being converted to a TIF file (in my programs TIF files are for viewing on a standard gamma 2.4 monitor after color correction for the most part) you can try making your DPX files. When you export the DPX files you need to take into account what the programs down stream will need as far as black and white clip points and gamma or curve adjustments. When you export CIN/DPX files from image data at normal gamma 2.22 image encoding for gamma 2.4 monitor correction, the output needs to be shifted by about gamma 1.7 and the black and white clip points be adjusted as will be needed, such as 95 for black and 685 for white as is common for film scans in order to convert to Cineon (tm) Log-C from Rec709 standard. 16bpc DPX files may not be as well supported as 10bpc DPX files, so you should make sure your downstream workflow programs all support 16bpc DPX files if you are going to export 16bpc DPX files. 16bpc DPX files are preferred to 10bpc DPX files for increased tonal range so that additional grading touch up does not introduce scaling errors that could introduce tone banding in your finished end use frames. There is also a probe in the Levels#1 and #2 control screens of the Grading command in DANCAD87.EXE (tm). The Key commands for DANCINEC.EXE (tm) probe are: P or p = toggle probe on or off, when on you should see a yellow rectangle over your graphics image display, the area within the probe is averaged to get the digital "density" reading. i = increase the probe increment value, wraps around from max. back to min. I = decrease the probe increment value, [i] increased, [Shift]+[i] decreases. h = increase probe height. H = decrease probe height. w = increase probe width W = decrease probe width + = makes probe rectangle area larger (moves slower to position on the image when larger with the cursor keys) - = makes probe rectangle area smaller (moves faster to position on the image when smaller with the cursor keys) 4 or [LeftArrow] = move the probe rectangle left on the image. 6 or [RightArrow] = move the probe rectangle right on the image. 8 or [UpArrow] = move the probe rectangle toward the top of the image. 2 or [DownArrow] = move the probe rectangle toward the bottom of the image. B or b = toggles the on screen text brightness so it is easier to read, or does not distract, press several times to rotate through the various mode options for the text brightness of the readouts. u = rotate increase through list of probe readout units types, these change the ratio of the screen 8 bit numbers to 10, 12, 16, or real number readouts etc. Keep pressing to see all the options display, it wraps around max. to min. again. U = rotate decrease through the list of probe readout units types, like [u] but going the other way, so if you bypass what you want you can "back up". ^X aka [Ctrl]+[X] = abort viewing (in case you are viewing a set of frames, otherwise press [Return] to see the next frame). Additional commands may be added in the Graphic viewer later. To get the Graphic viewer you press [T] for Test mode in the Execute sub-menu #3 from DANCINEC.EXE (tm)'s main menu. After the Graphic screen finishes displaying the "Test" processing using the currently selected process method you can press [P] or [p] to activate the digital density Probe. The sample area is not just one pixel since noise in the image would produce errors in the reading, so you should adjust the sample area within the yellow rectangle to be as large as you can while excluding parts of the image that are not the sample tone or color. That is if someone in the frame is holding an 18% Kodak (tm) gray card, you would "zoom" in the width and height of the yellow rectangle then center it over just the gray card to get a reading. If your reading is not close to the right value, like 0.462 for mid-tone, then one of these things may have gone wrong: 1) Your original exposure was not correct, too much exposure making the reading high and too little exposure making the reading low. 2) You did not select the right preset for your EI/ISO or K value etc. when you processed your images, or you did not use the probe in DANCAD87.EXE (tm) right when you make the KCC files using its Grading command if you are doing a full manual color balance on RAW data. 3) You have the input or output LUTs set wrong, CIN/DPX need gamma adjustment to screen gamma as well as black and white clip level adjustments set right. BMP files can have ITU/CCIR 601 limits set or not set to increase or reduce their range on loading or saving. If any of those LUT are on or off and so the workflow is incorrect, your mid-tone as well as other tones will not come out right. Things are simpler with TIF files which are always full range in my programs, so if you get confused try looking at a TIF conversion by changing the result file type. 4) If you are making DNG or BIN/RAW output, then there may be no display LUT used (because you did not select one to be used in the CIN/DNG I/O setup menu), and you will see a "linear" gamma 1.0 greenish display, that is what RAW data looks like on a standard gamma 2.4 monitor. To avoid seeing the data as RAW for DNG files you need to set the graphics read input LUT, clip, and gamma (which is in the graphics configuration #4 from the DANCINEC.EXE (tm) main menu, not in the process method setup, since that just changes how the data is displayed and does not change the data in the file itself). --- USING THE EXTERNAL VIEWER DANGUIVU.EXE (tm) AND ITS PROBE The external viewer program DANGUIVU.EXE (tm) works much like the full screen graphics viewers built into the 32bit DOS and Windows Console (tm) program versions, but lets the user use the mouse to move the probe around and to resize the probe, etc. like this: [MouseLeftButton] = Position the probe area around the spot pointed to by the mouse arrow cursor. If the probe has not been turned on by using the [P] key on the keyboard, clicking the [MouseLeftButton] while the mouse arrow is inside the graphics display area of the DANGUIVU.EXE (tm) window will turn the probe on and locate it centered on that point. If you hold the [MouseLeftButton] down you can drag the probe area around on the processed frame image and read the brightness values averaged from the probe area. Let up on the [MouseLeftButton] to fix the probe at that point. [MouseRightButton] = The [MouseRightButton] has more than one function depending on where the mouse arrow is. If you put the mouse arrow over the corner of the probe rectangle area and hold down the [MouseRightButton] then slowly move the mouse you can adjust the size and shape of the probe sample area. If you move the mouse arrow far away from the probe area and click the [MouseRightButton] you can turn off the probe so you can see the frame image without any displays on the screen. If the probe is off, and you press the [MouseRightButton] you can toggle through a list of color options for the readout display, like pressing the [B] key on the keyboard, you can then click the [MouseLeftButton] to turn the probe back on if you want to use it. [MouseCenterButton] = Quit the DANGUIVU.EXE (tm) program back to the caller program, like pressing [Esc] or [Return] of [SpaceBar] on the keyboard. There are some additional Keyboard key commands in DANGUIVU.EXE (tm): [T] = Tutor help screens showing the key and button commands. [B] = Four color options for the readout display rather than just dim and bright as setup in the VESA and DIRECT-X (tm) full screen graphics options. The reason white on black and yellow on black were added is that if you are using a 16:9 monitor to display a 16:9 image more or less, then there is less gray area on the graphics display for the readouts and the readouts will fall over the image area (if configured or toggled active). When you setup the config for DANGUIVU.EXE (tm) in the calling program, DANCINEC.EXE (tm) etc., you can set the maximum image width to the width of the desktop and the height to 13 pixels less than the height of the desktop. If you have a 1920x1200 pixel monitor, you can maybe get an image area of 1920x1187, so large enough for HD display 1920x1080 at 1:1 pixel maybe. On a 4:3 CRT monitor running the desktop at 2048x1536 you can get an image area of 2048x1523 which would be large enough for viewing 16:9 and 1.85:1 2K images at 1:1 pixel, maybe. You can set the window smaller, but its good to have some mid-tone gray above and below the image as a reference and for the readouts if you have the monitor and desktop size for some extra space in the DANGUIVU.EXE (tm) window image area. The target values to read a Kodak (tm) 18% Gray card shot in your images at the normal EI/ISO would be 0.462 for BMP and TIF files encoded gamma 2.2 for monitor viewing, and 10 bit code 470 for CIN or DPX files in Cineon (tm) Log-C 95-685 limits. Press [u] to change the units of readout display for the probe. If the Cineon (tm) files are processed through the monitor encoding gamma 2.2 LUT for viewing, then you would read close to 0.462 also. --- USING ANOTHER EXTERNAL VIEWER PROGRAM In addition to the DANGUIVU.EXE (tm) program you can use almost any other Windows (tm) graphics program that can take the name of a 24bpp BMP file and display it. You set this up in the graphics configuration in DANCINEC.EXE (tm), and the others, by entering the full path and filename of the viewer program, and the image size you want to export the "preview" image as. The BMP file that is exported to the image viewer program is pre-compensated for monitor viewing and is not meant to be used, saved, or edited unless you want to make notes on it to send it to someone as an example of color correction and such. This exported BMP file is run through monitor LUT to convert the values in the result frame made by the process method into something that looks on the monitor like it would more or less on the movie screen in the theatre. You can select to have the monitor LUT used or not in making the BMP file of the process method result file as you like with the graphics config setup. The reason that the result frame may need being run through a LUT is that if it is CIN or DPX type, it has image tones not at gamma 2.2 encoding for viewing, so conversion is needed to have the result file's image tones look and measure normal for a monitor viewable image. So DANCINEC.EXE (tm) makes a result frame in whatever file type you set it up to make, BMP, TIF, CIN, DPX etc. and then it converts that in to a common BMP file that is passed onto the viewing program with or without LUT adjustments as setup so you can judge the end result better. If you setup the converted BMP to be the same pixel size as the result frame files you want to view, then you reduce losses from the automatic resize that is otherwise applied on export. If the result frame size is larger than your desktop and you view the exported image made the same size as the result frame in your external viewer 1:1 pixel mode then you will not be able to see the whole image, but you will be able to inspect the result resolution and filtering easier. If you do view results 1:1 in an external program, and view close to your monitor you will see small processing artifacts that would be less visable to viewers after the images are transferred to 35mm movie film and projected or at least viewed at a distance of 3x the monitor diagonal. Monitors are in general sharper than projected images, and people would not be so close to the screen as you are with your noise almost touching the monitor faceplate. Another issue aside from artifacts or blur made by the resize in my programs to fit the result image into the graphics display BMP file, full screen or window, is that the program you are using to view with may resize or otherwise add its own artifacts such as banding to the images you see, so you need to be aware of any artifacts a particular image viewing program can introduce to what you see. Be sure to set the monitor correction to none in any external viewing program, and also set the desktop gamma, brightness, and contrast settings at default, since any adjustment to the Windows (tm) or external viewing program image adjustments can "de-calibrate" the display from its natural gamma 2.4 and so make what you see no longer relate to anything "normal" causing problems with color matching. The monitor you use for grading and color checking of your processed frames should be at native gamma 2.4 all the time, and so no adjustments in the viewing program or the Windows (tm) desktop would be needed or wanted. Attempts to "calibrate" your monitor may in fact be making it "de-calibrated" to all other normal monitors running at gamma 2.4 (images encode at gamma 2.22 for display on monitors at gamma 2.4 and DCP projectors at gamma 2.6, 2.6 is used because the theatre is darker than your grading room should be.) Standard image encoding for monitor viewing is gamma 2.22 which is really gamma 0.45, 1/2.22 = 0.45, 18% = 0.18^0.45=0.462, 0.462*255=118. Some LCD monitors are not showing at CRT monitor standard gamma 2.4, perhaps, you should avoid using such non-standard monitors in order to be able to tell what your images should look like. If there is a "probe" in your graphics program that displays the RGB values of a pixel, then in 8bit values the RGB for a Kodak 18% Gray card should be R=118 G=118 B=118. The 90% White side of the Gray card should be between 118 and 255, but NOT 255 for soft-clip images where there is a S- curve used for the highlight detail. You can display the same image using my Full Screen display, DANGUIVU.EXE (tm), and your graphics viewer of choice to see if the colors and image tones more or less match, then try other computers and monitors, the results should match for all of them, if not you need to find out why and fix the ones that don't match. If other people are using your computer monitor or display program be sure to tell them NOT to make any adjustments, tape over the controls, and check with a reference image each time you come back to your computer to make sure that they did not fiddle with something that changes the display of your reference image. The same goes for your audio monitor levels and EQ. --- MEANING OF COLOR BLOCKS NEXT TO PATCHES MADE BY PROCESS METHOD 95510 The special RGB to RGB process method 95510 "tags" CIN and DPX files with reference tone patches so you can use the digital "density" probe to try to make sense of what is going on with those file types since the normal Cineon (tm) aka Log-C type CIN and DPX files have mid-tone 470 shifted about gamma 1.7 within the 95 to 685 range which puts it above the normal 0.462 level for mid- tone encoding for image files of gamma 2.22 for viewing on standard monitors of gamma 2.4, and so you cannot look at such a CIN or DPX file without the image data being processed in LUT, clip points, and or gamma correction (if you have the data black clip at 95 and white clip at 685, if the black clip is 0 and the white clip is 1023 then mid-tone will be close to 0.458 to 0.462 but the image will look washed out and grayish because black is lifted quite a bit (that is the normal look of Cineon (tm) Log-C files without monitor correction)). You can pass the CIN or DPX data through DANCINEC.EXE (tm) without conversion to normal monitor gamma, and if you do not have the graphics preview's CIN/DPX input LUT on, you will be able to see and measure the tones in that CIN/DPX file more or less as they are (there is some rounding of the readings since the probe currently reads the screen buffer, not the file itself, but those small errors are not large enough to be an issue for filmmaking as the end projection will probably be off far more than +/- 0.00392 or +/- 0.392%. Since the CIN or DPX file can be large or small, I did not use text graphics burn in to mark the gray tone patches, rather I used small color squares next to the gray patches, in order to understand what those mean, I will give you a table here that shows you. The values are given in the 10bpc values, for 16bpc DPX files the values would be scaled in ratio. Gray = 470 (mid-tone 18% Kodak (tm) Gray card reference) Blue = under 95 (below black clip) Yellow = over 685 (over white clip) Cyan = 95 (black clip reference value) Magenta = 685 (White clip reference value) Green = 445 (Kodak (tm) D-Lad (tm) reference gray for prints density about 1.03) Red = 0 and 1023 (far limits of full range, max. black and white for full range) When you view the "tagged" CIN or DPX file you will be able to see gray patches OUTSIDE the 95 to 685 range, those are marked red then blue and yellow then red. You can see those "out of range" values when you view the "full" range without an input LUT being active. In its minimum operation the input LUT would cut off everything under 95 and over 685, so the patches under 95 marked blue would become black like the bottom box marked red, and the patches above 685 marked yellow would become white like the white box marked red (when soft clip is not in use). The reason red can be used to mark two colors, is that the lower end red is or should be always black and the upper end red is or should always be white, so you can tell them apart that way. The reason they need to be marked at all as red, is so you can tell the "full range" end patches from the other black and white patches when the clipping LUT is active, that is when the clipping is active the lower 0 black red tagged patch should be just as black as the Cyan tagged 95 patch, and at the upper end when clipping is active without softclip the Magenta 685 patch should match the 1023 Red tagged white patch. There is no "soft-clip" for the bottom of the tonal range, although you can use the CIN/DPX input KCC file if you want that. "Soft-clip" is used to pull down image data between 685 and 1023 to be under 685 so that you can see the "out- of-range" highlight detail when the CIN/DPX file is later used. "Soft-Clip" only acts on tones above 470 so that it does not darken mid-tone or reduce tones under mid-tone. I increased the range of "soft-clip" to do a "better" job of saving highlight detail with Digital Cinema camera images. For images from a film scanner, the film's soft shoulder would have already reduced the contrast in the higher values, but with digital images there may not be a "shoulder" to the upper tones, so my expanded range "soft-clip" can pull "all" tones above 685 down enough that you should be able to see them, although at lower contrast as they get closer to the maximum brightness values. The data code values for the gray patches on the gray steps added to your CIN/DPX file using process method 95510 are: Left side gray steps (in sound track area of image area 2048x1556 like Kodak (tm) D-Lad CIN image), patches Top to Bottom on left of image: 95 (tagged Cyan, black clip level, should match 0 next to it in normal use) 0 (tagged Red, lower limit, full range min.) 65 (tagged Blue, under black clip) 95 (tagged Cyan, black clip level) 125 (untagged, should be visible as almost black in normal use, black clip + 1/3 stop, there are 90 counts per f/ stop, so 95+(90/3) = 125) 180 (untagged, should be visible as dark gray in normal use, standard 2% black level for film scans) 445 (tagged Green, Kodak D-Lad mid-tone standard for print density about 1.03) 470 (tagged Gray, 18% Gray card standard for Cineon (tm) Log-C data files) 635 (untagged, lowest Kodak softclip break point, 685-50 = 635) 655 (untagged, white clip - 1/3 f/stop, 685-(90/3) = 655) 685 (tagged Magenta, white clip, about 90% white when soft clip is not used) 715 (tagged Yellow for over white clip, white clip + 1/3 stop, 685+(90/3)=715) 1023 (tagged Red, upper limit, full range max.) 685 (tagged Magenta, white clip level, should match 1023 next to it in normal use, unless soft clip is used to taper the highlights roll off) The Top gray step is 1/2 stop steps centered on 470 the value for an 18% gray card in the image, left to right: 95 (tagged Cyan, Black clip point, should match 0 patch in normal use) 0 (tagged Red, minimum value for image data, Max. Black, -5.22 stops from mid- tone 470) 20 (tagged Blue, under Black Clip, -5 stops from 470 mid-tone) 65 (tagged Blue, under Black Clip, -4.5 stops from 470 mid-tone) 110 (untagged, -4 stops from 470 mid-tone) 155 (untagged, -3.5 stops from 470 mid-tone) 200 (untagged, -3 stops from 470 mid-tone) 245 (untagged, -2.5 stops from 470 mid-tone) 290 (untagged, -2 stops from 470 mid-tone) 335 (untagged, -1.5 stops from 470 mid-tone) 380 (untagged, -1 stop from 470 mid-tone) 425 (untagged, -0.5 stop from 470 mid-tone) 470 (tagged Gray, 18% Kodak Gray Card Standard for film scans) 515 (untagged, +0.5 stop from 470 mid-tone) 560 (untagged, +1 stop from 470 mid-tone) 605 (untagged, +1.5 stops from 470 mid-tone) 650 (untagged, +2 stops from 470 mid-tone) 695 (tagged Yellow, over white clip, +2.5 stops from 470 mid-tone) 740 (tagged Yellow, over white clip, +3 stops from 470 mid-tone) 785 (tagged Yellow, over white clip, +3.5 stops from 470 mid-tone) 830 (tagged Yellow, over white clip, +4 stops from 470 mid-tone) 875 (tagged Yellow, over white clip, +4.5 stops from 470 mid-tone) 920 (tagged Yellow, over white clip, +5 stops from 470 mid-tone) 965 (tagged Yellow, over white clip, +5.5 stops from 470 mid-tone) 1010 (tagged Yellow, over white clip, +6 stops from 470 mid-tone) 1023 (tagged Red, maximum value for image data, +6.14 stops from 470 mid-tone) 685 (tagged Magenta, should match 1023 patch in normal use unless softclip is used to roll off the highlights, then you can see how much softclip is pulling down the 685 point relative to maximum white) The Bottom gray step is 1/2 stop steps centered on 445 the value for The Kodak (tm) D-LAD gray reference patch on the D-LAD image that corresponds to about density 1.03 on the film release print, left to right: 95 (tagged Cyan, black clip level, should match 0 patch in normal use as Black) 0 (tagged Red, minimum value for image data, -4.94 stops from 445 D-Lad reference level) 40 (tagged Blue, under black clip 95, -4.5 stops from 445 D-Lad reference) 85 (tagged Blue, under black clip 95, -4 stops from 445 D-Lad reference) 130 (untagged, -3.5 stops from 445 D-Lad reference) 175 (untagged, -3 stops from 445 D-Lad reference) 220 (untagged, -2.5 stops from 445 D-Lad reference) 265 (untagged, -2 stops from 445 D-Lad reference) 310 (untagged, -1.5 stops from 445 D-Lad reference) 355 (untagged, -1 stop from 445 D-Lad reference) 400 (untagged, -0.5 stop from 445 D-Lad reference) 445 (tagged Green, 445 is D-Lad image reference Gray for film print density about 1.03, not 18% Gray which is 470. The film LAD used 16% Gray for its LAD reference Gray patch, so both the digital and film LAD are lower than 18% Gray card on shots that are taken off movie negatives) 490 (untagged, +0.5 stop from 445 D-Lad reference) 535 (untagged, +1 stop from 445 D-Lad reference) 580 (untagged, +1.5 stops from 445 D-Lad reference) 625 (untagged, +2 stops from 445 D-Lad reference) 670 (untagged, +2.5 stops from 445 D-Lad reference) 715 (tagged Yellow, over white clip 685, +3 stops from 445 D-Lad reference) 760 (tagged Yellow, over white clip 685, +3.5 stops from 445 D-Lad reference) 805 (tagged Yellow, over white clip 685, +4 stops from 445 D-Lad reference) 850 (tagged Yellow, over white clip 685, +4.5 stops from 445 D-Lad reference) 895 (tagged Yellow, over white clip 685, +4.5 stops from 445 D-Lad reference) 940 (tagged Yellow, over white clip 685, +4.5 stops from 445 D-Lad reference) 985 (tagged Yellow, over white clip 685, +6. stops from 445 D-Lad reference) 1023 (tagged Red, maximum level for image data, full range white, +6.42 stops from 445 D-Lad reference) 685 (tagged Magenta, white clip level, should match 1023 patch beside it unless softclip is used then shows how much white clip has been lowered) There is also a "tone sweep" or "tone wedge" that has "all" image tones, the key Cineon (tm) tones are marked with thin lines above the "tone sweep" with the same colors as are the tone patches. You can use the "waveform" display option in DANCAD87.EXE (tm)'s Grading command's Chroma Matrix control screen to display the tone sweep as a curve like a curve on a graph so that you can see the effect of any tone adjustments you make to the "tagged" CIN/DPX file. You would then go back and use the KCC files made while grading the "tagged" CIN/DPX file for processing the untagged CIN/DPX files so as to not have the "tone sweep" in the image area of the result frames. (In the Grading chroma Matrix defaults to the vectorgraphic display, you need to press the key command and redraw the screen to see the waveform display of the loaded image.) Some information about the 10 bpc Cineon (tm) Log-C type and DPX files: 1) The data range that is displayed without softclip is the data between hard clip values 95 and 685. In 16bpc files the data values are about 64 times as large but in ratio to the 10bpc ones. That is 95x64=6080, and 685x64=43840. there is an issue with the full scale conversion of 10bpc to 16bpc, since (1023*64)=65472 and full scale 16bpc is 65535, so 65535-65472=63 error. So you may lose a little using 16bpc files going between programs that use 64x or 64.06158358x or 63.93847562x ratio for converting the 10bpc values to 16bpc. 2) Each f/ stop of image data in the scan is 90 values, so 1/2 stop is 45, and 1/3 stop is 30. 3) The normal hard clip range is 685-95 = 590/90 = 6.55 f/ stops of camera exposure. The actual range from the subject would be larger because the film has lower contrast in the foot and shoulder, that varies depending on the film stock and processing used. Same for Digital Cinema cameras since they use a curve to compensate for the linear sensor data having too much gain in the highlights and shadow areas of the image. If the camera negative was pull or push processed or the sensor gain was higher or lower the range of subject f/ stops could be larger or smaller than the 6.55 f/ stops of data range in the Cineon (tm) Log-C type CIN/DPX file. 4) The range between the 18% and 90% sides of the Kodak Gray card is about two and a third f/ stops, so about +225 in the CIN/DPX file, 470+225=695 which is a bit over white clip 685, if you take white clip as 90% = 685, as Kodak (tm) states, then 685-470=215 so 215/90=2.39 stops, so between 18% and 90% white clip is 2.39 stops. If you move mid-tone down to 445 then its 685-445=240 so 240/90=2.67 stops. 5) Film was not linear with respect to the change in the lens iris making exactly the same density change in the camera negative for each f/ stop that the lens was set to more open or closed, so actual camera f/ stops will not be exactly 90 counts in the CIN/DPX file. Also the S-curve of film, or the EI/ISO curve used in the Digital Cinema Camera, will make the counts per f/ stop smaller in the "foot" (shadow areas) and "shoulder" (highlight areas) of the exposure range. Force processed film negatives would use more than 90 counts per image f/ stop, and pull processed film negatives would use less than 90 counts per f/ stop. 6) The "printing density" formula is CV*0.002 where CV is the value in the CIN/DPX 10bpc file, so 90*0.002 = 0.18 density, but that is not one f/ stop since 1 f/ stop of density is 0.30, the reason is that the processed camera stock has a contrast lower than 1:1 so you get 0.18/0.30 = 0.6 as the contrast ratio between the camera negative and the Cineon (tm) data values. Prints are made higher than "life" contrast near mid-tone, the 0.6 camera negative contrast is combined with about 3.0 contrast in the print stock to end up with a contrast about 1.8x life, the higher contrast makes up for some lens flare and room and bounce light in the movie theatre which if there was no excess contrast in the print would make the projected image look a bit gray and washed out. The recorder intermediate film and a neutral 1:1 contrast near mid-tone but its contrast may be less than 1:1 in the foot or shoulder of its transfer curve, in in prints made the contrast is higher near mid-tone and less in the bright and dark parts of the image. If the intermediate film is push processed, then you would get more density change in the print for a given count in the CIN/DPX file, if the intermediate film is pull processed then you would get less density change (contrast) when printed to the print stock. Anyway, the assumption is that if you increase or decrease the CIN/DPX file's data one value, the light on the film recorder screen should act as though you added or subtracted a filter with a density of 0.002. Kodak (tm) must have thought that a "printing density" change of 0.002 was small enough that it would not cause tone banding in the final print, because the density change in the final print would be 3*0.002=0.006. Some kinds of release print stock could vary from contrast of 3x, and would depending on the processing time and temperature etc. The eye can see 0.025 density change, that is about 4x as large, CC filters smallest value I think was 0.025, like CC2.5M etc., so the 0.002 printing density step size does not have much extra density resolution to spare once you get to the high contrast release print, that is you may not see 0.002 brightness change with your eye looking at the film recorder's screen, but after the contrast increase of printing the density change will be larger and closer to visible. If you grade a 10bpc CIN/DPX file and increase the contrast, you can add to the contrast increase from what was in the source CIN/DPX (film scan etc.), so the increase to the contrast could be enough for you to see tone banding in the projected movie. Because of the banding issues with 10bpc CIN/DPX its better to use 16bpc DPX files to have extra tones to work with in grading (the step size is about 1/64th as large so you have about 64x as much grading contrast increase range before you see tone banding if the scanner or camera is 16bpc to start with). 7) With 590 tone steps between 95 and 685, you get a printing density range of 1.18 in the color intermediate film (0.002*590=1.18), so 1.18*3=3.54 in the color print, actually maybe a bit less depending on the maximum density range of the print stock used and the processing variations. Control of the contrast of the intermediate and print combined for the tones near mid-tone should be considered to be about 3x for reference (its less in the highlights and shadows because the film stocks both have S-curves). 8) Because there are more than 255 tone values in the CIN/DPX 10bpc file, in fact more than 511, being 590, a 9 bit ADC is not enough to display all of them on the film recorder screen, so a 10 bit video board would be needed. Dithering of the 590 tones through an 8bit video board could avoid some tone banding but would add some visible noise to the image, unless the film is exposed several times so that the dithering averages out (or the screen updated during exposure so that its dithered with different values +/- on each pixel during exposure). 9) The gamma of Cineon (tm) Log-C images is not at Rec709 HD encoding gamma 2.22 for viewing at monitor gamma 2.4 so additional gamma correction is needed to shift mid-tone to the right value, in the case of Rec709 HD encoding gamma 2.22 for gamma 2.4 monitors mid-tone is about 0.462 of 0 to 1 range. CIN/DPX value of 470 in the range 1023 is 0.459, so Kodak (tm) seems to have located 18% Gray at a point that would "stand still" if you view the CIN/DPX file with or without the clip points and anti-Log-C monitor gamma or tone correction LUT, perhaps. 10) When printing film-to-film the film stock being printed to has an S-curve, or at least it did when there was more silver in the film, so the result is not the same as the hard-clip "window" used in the Cineon (tm) concept, where grading was to have been done by adding or subtracting values to the Red, Green, or Blue values in the CIN/DPX file. Sliding the tones up and down with a "brightness" shift on the hard-clip points 95 and 685 is not like film-to- film printing where the contrast varies with exposure of the film bring printed to. So Grading should be done with software tools that allow for curve adjustments, rather than by adding or subtracting "brightness" density values to the CIN/DPX data. 11) The full negative density range of the Cineon (tm) scanner was said to be 2.048 density range, 0.002*1024=2.048. Although the scanner light would be adjusted so that base density of the negative, like in the frame lines, would be just under CIN/DPX value 95 so the range was less in use, 0.002*(1023- 95)=1.856 about. Anything on the normal exposed negative over density 1.856 above base would be white clipped, although there seems to have been a provision to shift the 95 black level's exposure lower in the scanner for over- exposed negatives (putting base density lower than 95). Since there are 90 points per f/ stop, putting the black level at 5 rather than 95 would allow for negatives that were overexposed. You don't need to go much lower than -90 from the 95 level for black since the shoulder of the negative would not hold 90 extra points for each additional stop of over exposure do to the S-curve of motion picture negative film stocks. 12) Because Kodak (tm) states that the "printing density" is the same as the scanner density per file value, 0.002 in both cases, that means that they intended they film recorder to mimic the scanner brightness change on a 1:1 ratio, that is the same on both. If you build your own DIY film scanner or film recorder, you can calibrate them knowing that the "density" change should be 0.002 per Cineon (tm) point for 10bpp files, if yours is higher you reduce the gain or contrast adjustment, if less you increase the gain or contrast adjustment. As noted above, that gives you a light meter range of (0.6*6.55) = 3.93 f/ stops between black and white on the film recorder screen, or what the digital camera must capture, the 0.6 is the difference between camera negative contrast and the "real world" that the light meter reads, and 6.55 is the range of f/ stops in the camera negative's exposure (subject brightness range) in the "real world", and 3.93 is the film recorder screen's brightness range since you are simulating the brightness range of the camera negative and not the "real world", that is taken form the 90 points per f/ stop of subject contrast and the range of 95 to 685 of the Cineon (tm) black to white range, 685- 95=590/90=6.55, then 90*0.002=0.18 density per f/ stop on the negative, then since 0.30 is the density change in the "real world" per f/ stop, you get 0.18/0.30=0.6, so you reduce the film recorder brightness range to 6.55*0.6 and get a light meter reading of 3.93 f/ stops from the film recorder screen at "black level" (which may not be 100% black because you need some slight exposure to bring the print or intermediate density into the useful range) and the "white level" When you check your intermediate you should read a density change of about 6.55*90*0.002=1.179 between the base density plus fog and flare in the image area and the D-Max representing the black to white range from the CIN/DPX file's image data (since those are digital hard clip points there is no density outside that range like there would be in film-to-film printing). 13) ARRI (tm) has a chart on their web site that gives these values for density in the negative vs. the CIN/DPX file's value: 95 = 0.00 density above base density = black calibration level 180 = 0.17 density above base density = 2% black subject 470 = 0.75 density above base density = 18% gray subject 685 = 1.18 density above base density = 90% white subject 900 = 1.61 density above base density = D-Max on negative (would vary) 1023 = 1.86 density above base density = scanner white clip 14) ARRI (tm) also gives some formulas about CIN/DPX files: DPX_code_value = 95 + (density_above_base * 500) so 470 = 95 + (0.75*500) density_above_base = (DPX_code_value/500)-0.19 they get: 0.19 = (95*0.002) since each DPX code value is a 0.002 density step. 15) The DPX Log files in Cineon (tm) form assume: The negative was processed to gamma 0.6 so you get 90 0.002 steps per f/ stop of negative exposure. 90*0.002 = 0.18 density, In the "real world" on f/ stop is 0.30 density, the real world would be gamma 1.0, 0.18/0.30 = 0.6 the gamma of the negative being scanned. Note that release prints are processed to a contrast higher than would produce a density change equal to real world, so you will NOT read 0.30 density change per f/ stop on the release print or holding a light meter at the movie screen. The contrast of the negative is 0.6 and the Print stock 3.0, so you get 0.6*3.0=1.8x real world, so if you hold a light meter on the movie screen and shoot an 18% Kodak (tm) Gray card, and you make three shots, one +1 f/ stop, one at exposure, and another -1 f/ stop, you should see an 1.8 f/ stop change on the light meter or if you measure the film in a densitometer you would see a density change of 1.8*0.30=0.54 between the center and the higher and lower frames. Because film has an S-curve, going more than +/- 1 f/ stop will produce lower changes in density and lower changes in the light meter reading. The maximum change is near mid-tone. Because mid-tone should be about 1.03 density or a bit thinner in a release print since the D-LAD and LAD are darker than mid-tone, you would measure an absolute density of about 1.03+0.54=1.57 for the darker frame and 1.03-0.54=0.49 for the lighter frame. If you read a spread larger than that you have your system contrast too high, it will probably read a little smaller density spread if you have things calibrated right. Processing and scanner "gain" adjustments can change the relationship between f/ stops at the time of exposure and the DPX counts per f/ stop. But if you are building your DIY film scanner and want your DPX files to turn out right, then you want to Grade your RAW scan files so that you come out with + or - one f/ stop at the time of shooting being 90 DPX code values. To calibrate that (for use with DANCINES.EXE (tm) use in making your own scanner) you would shoot an 18% Gray card at the films stated EI/ISO speed using a light meter that is calibrated to take a reflected light reading off 18% Gray cards, and have that color negative processed normal, no push or pull processing and no bleach bypass. You would make a series of full frame 18% gray card exposures at correct exposure, then +1, +2, +3, +4, +5 stops and -1, -2, -3, -4, -5 stops using the lens iris and ND filters. You should not expect to measure perfect 90 count steps all the way along the film's transfer curve, but if you do it several times you can scan that film and average the readings to get your overall Grading curve to give about the right results, that is from -2 to +2 stops you should end up with about 4*90 = 360 counts in the CIN/DPX file made etc. If the negative is not 0.6 gamma, you can compensate somewhat with your grading of the RAW files from the camera in your DIY movie film scanner, that is you make a special WBC KCC file in DANCAD87.EXE (tm)'s Grading command and adjust the curve slope with the Curves#1 or #2 control screens. 16) ARRI (tm) gives a formula to convert DPX density values to TIF "monitor" gamma, relative_luminance = 10^((DPX_code_value-white_point)*0.002*negative_gamma) The white_point is 685 or 90% white The negative gamma is 0.6 That puts white_point as 1.0 and then its scaled with 4095 so that there are 4 stops of "headroom" for highlight detail, so 470 in the DPX file becomes, TIF_value = (10^(470-685)*0.002*0.6))*4095 TIF_value = (10^((-215)*0.0012))*4095 TIF_value = (10^(-0.258))*4095 TIF_value = 2261 2261/4095 = 0.552 - 0.458 = 0.094, 95/1023=0.093 ? As you can see, the DPX 95 black level in the TIF file is not being lowered to zero, it works out to 802 in the TIF. That results is a very dark TIF file, additional gamma correction or soft-clip curves would be needed to make a viewable result on the movie screen. 17) ARRI (tm) shows that the 0.002 density per step is 1/500 Log D 18) ARRI (tm) shows that densities are shifted 0.19 Log D (base density calibrated at code 95, 95*0.002 = 0.19) 19) ARRI (tm) has a table that shows the camera exposure vs. the DPX values: 95 = Base Density = -4 2/3 stops 180 = 2% Black = - 3 stops 470 = 18% Gray = +/- 0 stops 680 = 90% white = +2 1/3 stops 900 = D-MAX = +4 2/3 stops The D-MAX is for normal exposed negatives, the density can overrange clip at 1023 if the negative is overexposed, force processed, bleach-bypassed, or was of a high contrast range subject. For high contrast range subjects etc. the negative needs to be "pull processed" or the scanner gain reduced to avoid overrange clip at 1023. How you get those I figure without the film's S-curve effects, 95 = (95-470)/90 = -4.16 stops 180 = (180-470)/90 = -3.22 stops 470 = (470-470)/90 = 0 stops 680 = (680-470)/90 = +2.33 stops 900 = (900-470)/90 = +4.77 stops Those are SUBJECT f/ stops in the real world, if you want to check the light through the DIY scanner film gate, or the light off the film recorder screen you would multiply those by 0.6 so you get, 95 = ((95-470)/90)*0.6 = -2.5 stops 180 = ((180-470)/90)*0.6 = -1.932 stops 470 = ((470-470)/90)*0.6 = 0 stops 680 = ((680-470)/90)*0.6 = +1.398 stops 900 = ((900-470)/90)*.0.6 = +2.862 stops You would ONLY read the 900 code's value in the scanner only since its outside the film recorder's range (with softclip you can bring it into the range, but that is not what you want to measure here). On the movie screen you should measure these values because the release print is 1.8x the real world f/ stop change, (you will read lower values because the print stock has an S-curve though, lens flare, and lights on and reflected light in the movie theatre), 95 = ((95-470)/90)*1.8 = -7.488 stops (would measure maybe -2.5 to -3.5) 180 = ((180-470)/90)*1.8 = -5.796 stops (would measure maybe -2 to -2.5) 470 = ((470-470)/90)*1.8 = 0 stops 680 = ((680-470)/90)*1.8 = +4.194 stops (would measure maybe +2 to +3) There can be no 900 code value since the film recorder hard clips the range to be within 95 to 685, if you soft clip the highlights 900 would be lowered to 685 so 685 is always the limit for the film recorder. 20) ARRI (tm) implies that the Bell and Howell C printer has 50 light steps of 0.025 Log E each. 0.025/0.002=12.5 Cineon (tm) steps per Bell and Howell light house timing step. So to convert a timing light card or tape you multiply the values by 12.5 and that gives you the Cineon (tm) value to add or subtract from mid-light. In other words if mid-light is 25 and the tape says use light 32 then you get (32-25)=7*12.5=88, so you add 88 to the Cineon (tm) target value for grading, if you had a gray card at 470 then you would measure 470+88=558 in the graded CIN/DPX file. (You can "tag" any CIN/DPX file with the special process method that puts the gray steps and patched on your image test frames so you don't need a shot with a gray card to give you a patch to put the test probe on in DANCINEC.EXE (tm) and DANCAD87.EXE (tm)). The reason you would use the probe to check the tagged patch is that you would increase the mid-tone on a curve so as not to clip the highlights and you would probably want to pull down the foot of the curve so you do not get washed out blacks from grading up, so you are doing more than just changing the brightness by adding if you want better results by using a curve to simulate the S-curve to S-curve effect of film-to-film printing using my programs Grading commands and image processing LUTs. 21) ARRI (tm) seems to state that the mid-tone gamma of release prints is gamma 1.8 over the original subject contrast, that get that by 0.6*3.0=1.8, where 0.6 is the camera negative gamma 0.6, and 3.0 is the combined contrast (gamma) of the intermediate film exposed in the scanner, which is gamma 1.0, and the release print stock at gamma 3.0, so 1.0*3.0 = 3.0, but that would depend on the lab used and the current film stocks. The scanner and the film recorder are to be set so that they do not alter the negative contrast, that is the intermediate would show the same density as the original negative no matter if it was printed on a film printer or in a film recorder (not an easy thing to do, you need to adjust the LUT used to convert the scanner data and setup the film recorder LUT then double check the intermediate with a densitometer, it can take weeks and many tries to get close to right). Without correcting for the S-curve of the intermediate film stock with an LUT, the contrast would be a bit higher at mid-tone than 1.0 and lower than mid-tone near black and white clip, that is the intermediate would be processed a bit higher contrast than needed to compensate for lens flare in the film recorder, otherwise there would be an overall loss of contrast between the tones in the camera negative and the intermediate negative made in the film recorder. 22) It seems ARRI (tm) says that the scanner is calibrated so that the film grain is all above DPX code 95 so that the noise is not cut in half at black clip. I've gone into the CIN/DPX issues here because you will need to use process method 95510 to "tag" a copy of one of your keyframes in order to help adjust the CIN/DPX input and CIN/DPX output LUT and other settings i.e. black clip value, white clip value, and conversion gamma, so that you can adjust mid-tone to fall on the right value in your graded image frames. See the notes above in the DPX/CIN input and output LUT discussion section about using gamma 1.7 with DANCINEC.EXE (tm)'s input and output DPX/CIN LUT in order to convert "Log" DPX/CIN to HD Rec709 image encoding gamma 2.22 for monitor gamma 2.4 so you can look at how your images will look on the movie screen and TV set or computer monitor, and then also to convert monitor HD Rec709 encoded gamma 2.22 images to "Log" (Log-C or Cineon (tm)) DPX/CIN files for use in workflows that require those. As motion picture post production moves away from film prints and shooting film, the use of film Density based CIN/DPX files may become some kind of legacy issue that filmmakers will still need to deal with because much of the workflow processing has developed using the Density based code values. So a good grasp of how and why such Density based file encoding was used is still something filmmakers should understand, so I have tried here to give an overview that you can apply to your Digital Cinema Camera exposure and Post Production as well. --- SUB-TITLES DOT COMMAND FILES DANCINEC.EXE (tm) et al. has a provision in its Burn-in ([B] from Execute sub- menu #3 from the main menu) menu to have automatic sub-titles burned in to the processed RGB output frames. Here is some information on that subject taken from the v0.01 help screens in the Burn-in setup menus. As there may be changes since this document was written you should check the program version you are using for any changes or additional information on the subject of the sub-titles and burn-in. Because the sub-titles are done at the time of v0.01 release using the fixed symbol cell fonts files, you may want to use the active image doubler before superimposing the sub-titles, such as going from 2K to 4K, and then reduce the frames back to 2K or HD size to get some smoother edges to the text. You can also adjust the enlargement and reduction to adjust the size of the symbols with finer control than just the 2x, 3x and so on integer size multiplier in the Burn-in sub-menu. I may add other options for anti-aliased fonts later, but as DCP support their own sub-titles there may not be a real need for anti- aliased fonts here as they would mostly be used for workprinting and for making 35mm film prints in the few remaining years that film will be available. The title or sub-title text for title elements option #4 is named TITLE4.TXT and you put them in the root shot folders so their name and path would be like: C:\P0001\S0123\TITLE4.TXT You can use a text editor like NOTEPAD.EXE (tm) to write or edit the many TITLE4.TXT files needed. Since each shot would need its own TITLE4.TXT file you might need to make a thousand or more, and each once can have the same dialog or text in several languages such that you disable the burn-in element numbers in the #4 setup for the languages you do not want burned in a that time, but you can enable those elements and then disable the others to burn-in another language later. Or you can export all the edited frames to another project and treat all the shots as a single shot to make just one long TITLE4.TXT file for all the sub-titles or titles, but that may work slower as the text search for the "ins and outs" for each title text would have to scan the longer TITLE4.TXT file. To span cuts you can have the TITLE4.TXT repeat the text from the previous shot, but have the range start at the next shots first frame. The frame ranges don't need to be exact you can make the first 0 and the last 99999999 if you want the range to start on the first frame used, or to go to the last frame used. Press [Return] to go to the next info page. Because of the way this document was formatted the dot commands are not show aligned with the left margin of the page as they would be in actual use. When you make the TITLE4.TXT files be sure the dot for each dot command is on the left most character column and that you do not start non dot command lines with a period. Ignore the underline at the start of each dot command line, you would not insert those in use, they are just there so the HTML page for this document will format. The values used for each dot command are on the next TXT line. The dot commands must put the dot as the first char on each line. All values etc. are on the line after the dot commands: _ .e = select the title #4 burn-in element number 1 to 12. _ .o = select the order for the frames source or result numbers. _ .r = select the range of frame numbers to show this title or sub-title. _ .m = override typeface mode for char cell size font type (.m always before .f). _ .f = override typeface font filename for typeface mode (.f size must match .m). _ .s = override the char cell scale factor to enlarge or reduce the size. _ .x = override locate at this x axis char cell on the image. _ .y = override locate at this y axis char cell on the image. _ .l = override x and y axis char cell location on the image. _ .c = override color attributes for foreground, background, and shadow. _ .h = override typeface enhancement mode. _ .k = override kludge values. _ .t = text on single line to display. _ .b = block of text to display on several lines. _ .. = end marker for block of text on line below the last line of text. _ .i = import text from another file, filename of text source line or block. _ ./ = comment line, no function just for notes in TITLE4.TXT files. _ .z = end of dot commands to flag stop looking in the TITLE4.TXT file. Dot command .e usage: _ .e _ number_of_burn_in_element_1_to_12 Example: _ .e _ 12 Notes: The .e command must come before all others so that the others change the settings that correspond to the right burn-in title. The burn-in element numbers range 1 to 12. The burn-in element must be enabled before from the menus to work. Dot command .o usage: _ .o _ frame_number_source_or_result_code Example: _ .o _ 2 Notes: The .o command must come after .e and before all others so that; the others change the settings that correspond to the right burn-in title. The source frame can be 1=source frame number or 2=result frame number. The source/result frame numbers are the same for most but not all of the many process methods. Dot command .r usage: _ .r _ first_frame_number_in_shot_to_burn_in last_frame_number_in_shot_to_burn_in Example: _ .r _ 238 517 Notes: The .r command must come after .e and .o and before all others so that the others change the settings that correspond to the right burn-in title. The first frame can be 0 or n, and the last frame can be first to 99999999. The default range if you don't use .r is 0 to 99999999 so that the (sub) title will burn-in on all the frames of the shot by default. If you have a long shot you can use several .r commands to set ranges for two or more (sub) titles using the same burn-in element, just be sure that the .r command comes before the .t Dot command .m usage: _ .m _ typeface_mode_code_1_to_6 Example: _ .m _ 6 Notes: The .m command must come after .e .o .r and before .f for each burn-in element so any change to the settings can correspond to the right burn-in title. The typeface mode codes range from 1 to 6 for small to large char cell. You can also adjust the size with the .s scale factor, or by picking the burn-in option in the menu that selects before or after resize. If you adjust the size of the image before resize by making it larger and use the largest char cell, the results will probably look better for finished 4K or 2K frames for film transfer or DCI projection. Typeface_mode changes the char cell size and so affects maximum x & y. Dot command .f usage: _ .f _ typeface_filename Example: _ .f _ F64X112A.814 Notes: The .f command must come after .e .o .r and .m for each burn-in element so any change to the settings can correspond to the right burn-in title. The typeface mode codes range from 1 to 6 for small to large char cell so you need to use this to set a filename for the screen font file that matches the typeface mode you are using. Normally the typeface would be setup as the default in the setup menus, so you would not need to include this in the dot command files unless you need to change to another typeface for some other language etc. The raster typeface filenames supplied all have the file extension *.814 and are named like 5X8A.814, 8X8A.814, 8X14A.814, 16X28A.814, 32X56A.814, 64X112A.814, 64X112B.814 and so on. You can manually edit them in a text editor by converting them into *.F14 files and then back into *.814 files for use, see the commands in the [B] for Burn-in sub-menu of the Execute sub-menu in DANCINEC.EXE et al. You can also convert *.FON vector font files drawn in the CAD programs into the larger size *.814 files with the commands in the Burn-in sub-menu. The font size MUST match the size of the *.814 filename selected or the text will not Burn-in correctly. The .m command must come before the .f command so that the size is set before the filename selected, doing it the other way around will cause the .f filename to be replaced with he default *.814 file for the size selected. Dot command .s usage: _ .s _ x_scale_factor y_scale_factor Example: _ .s _ 2 4 Notes: The .s command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The scale factors increase the char cell size and so affect the x and y char location maximum values. Its best to use the largest typeface mode before using the scale factors to magnify the smaller typeface as that makes the char look "blocky". If you are inserting the burn-in before re-size and you have large frames then you may need to scale the largest typeface up to be large enough. You can set the first char locate values to -1 to auto-center the text. Press [Return] to go to the next info page. Dot command .x usage: _ .x _ x_char_location Example: _ .x _ 36 Notes: The .x command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The typeface mode and scale factors increase the char cell size and so affect the x and y char location maximum values. If you have the start for the text located too far right and the text length would put some of it off the right side of the image you may get a fatal error during execution of the process methods used, so you need to work out the text formatting carefully for your image size. Dot command .y usage: _ .y _ y_char_location Example: _ .y _ 54 Notes: The .y command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The typeface mode and scale factors increase the char cell size and so affect the x and y char location maximum values. If you have the start for the text located too far down and the text length would put some of it off the bottom of the image you may get a fatal error during execution of the process methods used, so you need to work out the text formatting carefully for your image size. Dot command .l usage: _ .l _ x_char_location y_char_location Example: _ .l _ 36 54 Notes: The .l (.L) command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The typeface mode and scale factors increase the char cell size and so affect the x and y char location maximum values. If you have the start for the text located too far over and the text length would put some of it outside of the image you may get a fatal error during execution of the process methods used, so you need to work out the text formatting carefully for your image size. Dot command .c usage: _ .c _ Rfg Gfg Bfg Rbg Gbg Bbg Rsh Gsh Bsh fg_tans bg_trans sh_trans Example: _ .c _ 1.0 1.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.462 0.0 Notes: The .c command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The fg values are for "foreground" or the color of the letters, numbers, and symbols, the bg values are for "background" and are the unused. parts of the char cell space, and sh are the shadow or outline color. The trans values are the transparency 0.0 is opaque and 1.0 is clear so the image shows through without anything blocking it. Normally you would set the foreground and shadow/outline opaque and the background part way or all the way transparent. Dot command .h usage: _ .h _ enhancement_code Example: _ .h _ 15 Notes: The .h command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. These are the enhancement options: 0=None. 1 to 9=Drop shadow, 1=thin to 9=heavy. 11 to 19=Outline, 11=thin to 19=heavy. Dot command .k usage: _ .k _ kludge1 kludge2 Example: _ .k _ 0 0 Notes: The .k command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. This is just a place holder now, as these values are not in use here. Dot command .t usage: _ .t _ title_text_goes_here... Example: _ .t _ What do you mean the corroborator needs cleaning? Notes: The .t command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. This command is used for titles and sub-titles that are only on one line of text, for multi- line titles use the .b command. If you use same text often, you can load it from another file using the .i command. The text is always on the next line under the .t command.'); Dot command .b and .. usage: _ .b _ first_line_of_title_text_goes_here. _ next_line(s)_of_title_text_goes_here. _ .. Example: _ .b _ When do you expect the car repair to be finished, I have to be in Denver by 3:10 today no matter what? Notes: The .b command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. This command is used for titles and sub-titles that are more than one line of text. The .. command must be on the line under the last line. Dot command .i usage: _ .i _ external_text_filename Example: _ .i _ C:\TITLES\SELECTED\SOMEFILE.TXT Notes: The .i command must come after .e .o and .r for each burn-in element so any change to the settings can correspond to the right burn-in title. The .i command would be used if you have some "stock" text that you need to burn in more than once, or if you need to generate text that would be read by dot command files in such a way that you change the text file without needing to change the dot command files, maybe for special multi-language use where you copy the folder to select which one is used by the dot command files, or any other use you need. Dot command ./ usage: _ ./ _ comment_text_on_one_line Example: _ ./ _ This is an example comment in a dot command file for titles. Notes: The ./ command must before the .z command. Comments can come between other dot commands for your own reference, they do not control anything or display. Dot command .z usage: _ .z Example: _ .z Notes: The .z command must come after all other dot commands since it ends the reading of the TITLE4.TXT file so the program then burns-in the title(s) or sub- title(s) into the current frame being processed.'); Each TITLE4.TXT dot command file goes into the shot folder that you want it's titles to show on the frames in that shot only. To run a title over edits you can repeat the title in the next shot's folder with the range set so that there is no gap. Example TITLE4.TXT dot command file text: _ .e _ 1 _ .l _ -1 -2 _ .r _ 35 278 _ What do you mean there are no bananas today! _ .r _ 523 715 _ .b _ The pineapples are very sweet today, _ why don't you get some of those? _ .. _ .z The text elements can be located manually by entering a X and Y char cell on the image numbered X=1 Y=1 for the upper left to X=X_Max Y=Y_Max Where: X_Max = width_of_the_image_in_pixels / width_of_the_char_typeface_in_pixels Y_Max = height_of_the_image_in_pixels / height_of_the_char_typeface_in_pixels You need to keep in mind if the text element will be burned-in before or after resize since the image size can change. You select the size of the typeface font to use from 5x8 to 64x112 etc. One issue with manual adjustment is that you need to do some math to be sure you have the text elements (including things like SMPTE and TOD) in the right spot on the screen and to adjust for text strings that can vary in length. To get around that issue, there are some "automatic" locate X Y options to automatically center or place text elements that compensate for the image resolution and text string length in some ways. Since the char cell is fixed size for each font, you might adjust the image size before resize to the target size to get fine adjustment if needed, write('using the brun-in insert before resize option. There are some "simple" auto-location options, and in additional text below some other that give additional range to the placement. For X placement: 0=X is (0.25*X_max_char) - (0.5*text_string_length) -1=X is (0.50*X_max_char) - (0.5*text_string_length) -2=X is (0.75*X_max_char) - (0.5*text_string_length) For Y placement:'); 0=Y is (0.20*Y_max_char) - (0.5*number_of_text_element_lines) -1=Y is (0.50*Y_max_char) - (0.5*number_of_text_element_lines) -2=Y is (0.80*Y_max_char) - (0.5*number_of_text_element_lines) So to center a text element like a title you can use X=-1 Y=-1. To locate the SMPTE readout under the subject you can use X=-1 Y=-2. If you have three SMPTE readouts you can use X=0 for the first, X=-1 for the second, and X=-2 for the third, and Ys to Y=-2. To put three more readouts over the subject change the Ys to Y=0. X and Y location values between -1100 and -1000 do this: For X placement: You invert Y then subtract 1000 and get 0 to 100 then times 0.01 to get n = 0.00 to 1.00. Adjustment in 100 parts. X = (X_max_char*n) - (text_string_length/2)'); So -1000=top of image and -1100=bottom of the image, -1050=center. For Y placement: You invert Y then subtract 1000 and get 0 to 100 then times 0.01 to get n = 0.00 to 1.00. Adjustment in 100 parts. Y = (Y_max_char*n) - (lines_of_text_in_element/2) So -1000=left of image and -1100=right of the image, -1050=center. X and Y location values between -2100 and -2000 do this: Location is toward the center away from all the edges. The value of n is found as in -1100 to -1000. gotoxy(1, 7);write('For X placement: if n < 0.5 then X = (X_max_char*n) if n = 0.5 then X = (X_max_char*n) - (text_string_length/2) if n > 0.5 then X = (X_max_char*n) - (text_string_length-1) For Y placement: if n < 0.5 then Y = (Y_max_char*n) if n = 0.5 then Y = (Y_max_char*n) - (lines_of_text_in_element/2) if n > 0.5 then Y = (Y_max_char*n) - (lines_of_text_in_element-1) X and Y location values between -3100 and -3000 do this: Location is always right or down from location. The value of n is found as in -1100 to -1000. For X placement: X = (X_max_char*n) For Y placement: Y = (Y_max_char*n) X and Y location values between -4100 and -4000 do this: Location is always left or up from location. The value of n is found as in -1100 to -1000. For X placement: X = (X_max_char*n) - (text_string_length-1) For Y placement: Y = (Y_max_char*n) - (lines_of_text_in_element-1) Be sure to check the current help screens in the Burn-in sub-menu to see if any additional options have been added since this document was written. --- MISCELLANEOUS The Windows GUI (tm) version, DANCINEW.EXE (tm), should not be used unless the Windows Console (tm) version, DANCINEC.EXE (tm), or the 32bit DOS version, DANCINED.EXE (tm), cannot be used since it seems to have more issues than the other two versions. All three versions have the same basic 80x25 text interface and source code in the first release, but the GUI version only uses the graphics display program DANGUIVU.EXE (tm) for the digital density probe feature. As you may notice the program functions become more problematic as you move from DOS to GUI and with newer versions of Windows (tm), that is not an accident, Microsoft (tm) has a "road-map" that makes problems for existing software and then they seem to refuse to allow anyone to fix the "bugs" they introduce. I tried talking with them, and you know where that went. Anyway, I hope you can maybe get at least one of the versions running so you can process frames, you might need to purchase some older "junk" computers to get things to run. If the 32bit version of FreeDOS (tm) gets to a point that it can be used, then you would have an alternative, perhaps. I plan on looking into Ubuntu (tm) Linux (tm), but so far I also have problems getting the graphics to work under it, although it might be possible to do batch frame processing without that working, using just Console display. I hope to look into portting the programs to work better with current Windows (tm) versions as I am able, perhaps, as well as looking into portting the code to run on MAC and or Linux, but I have no time table for that code port as I do not know if it is possible to get all the needed program functions working as they do in the DOS versions because there are no equivalent functions and much of the code would need to be adapted to work in different ways, if it can be made to, or at least if I can figure out a way for it to work well. Right now you best option seems to be to use Windows ME (tm) to run most of my programs. Using USB drives at the same time as DANCAD3D.EXE (tm) or DANCAD87.EXE (tm) may cause a system freeze under Windows XP Home (tm). This is not a normal program crash, the OS may go dead, perhaps due to a Windows (tm) "bug". A CRT monitor is more flexible since it can display many resolutions and aspect ratios in the Graphics Test display. For people forced to use LCD monitors I wrote DANGUIVU.EXE (tm) since it can be adjusted to have its graphics window fit just about any monitor shape needed (see the graphics config in DANCINEC.EXE (tm), as well as DANCINED.EXE (tm) and DANCINEW.EXE (tm)). It seems the ITU has set a gamma 2.4 standard for LCD monitors, but that is new so many monitors and HD TV sets may display at other than gamma 2.4 which may result in some confusion about the Grading of your images. To avoid having your graded images too light or dark, or off neutral color, use the probes in DANCAD87.EXE (tm)'s Grading command control screens, and in DANCINEC.EXE (tm) and DANCINED.EXE (tm) or DANGUIVU.EXE (tm) to check the actual data values in your images rather than to depend on how the images "look" on the monitor to your eyes. Grading "by the numbers" makes your reference "monitor independent". You should black out the windows in your Grading room so that the color of the light in the room does not vary over the course of the day or with the weather. The lights in the Grading room should be adjusted to normal levels and be color balanced so that your eyes do not get adjusted to some odd K value and make you adjust all your shots off color. Some people may paint the walls in their Grading room neutral 18% Gray. Once the dimmers on the lights are adjusted (you may need to add CTB gels on the lights to compensate for the dimmers), and the monitor setup and calibrated you should tape over the controls so that people do not fiddle with the settings without your knowledge and result in a mismatch between work done before and after their meddling. The same goes with the audio monitor level, it should be fixed using a sound level meter and never adjusted, if you want to have silence or reduced volume at times, do that by switching the sound from the control level control, to a second level control, otherwise you cannot get the levels to match since compressed audio needs to have its loudness adjusted by ear and not on a VU meter since the peak to average ratio is changed and a peak reading VU meter will not show the volume as it sounds to the audience's ears. You may need CTB gel over the room lights in the Grading room to match the light of the movie projector Xenon light. Hold an 18% Gray card in front of your monitor, and make a file with R=118, G=118, B=118 in MSPaint.exe (tm) and display that, then adjust the room lights and your monitor so that the gray on the monitor screen matches the gray on the 18% Gray card. The video board color and gamma adjustments should be set to default or neutral as should the monitor corrections in any graphics programs you use to look at the result frames. The Graphics Test display in my programs does not normally work through the Windows (tm) color adjustments (at least in the DOS version) and should display as the monitor is setup. If you start changing the "natural" white balance, color adjustments, and gamma adjustments in various places like the Windows (tm) desktop adjustments, application program adjustments, and the monitor adjustments, then you can introduce bias that would make you think the result files look other than what they do. My programs are intended for a neutral adjusted gamma 2.4 CRT monitor of high quality made for Grading. Anything you adjust that deviates from the "natural" CRT transfer curves of the monitor you are using for Grading would cause confusion about the result files "look" and so be avoided. In general the programs should not crash in normal use, if you get an error message like 216 or 203 on the screen, then the program probably ran out of memory. If your computer system freezes so that [Ctrl]-[Alt]-[Del] does not break you out of the program, then its probably a Windows (tm) issue since in protected memory mode, you should always be able to press [Ctrl]-[Alt]-[Del] to break out of the program if you need to. Its best to never press the "Windows" (tm) key when you are in my program's graphics display since Windows (tm) may not be able to restore the programs, sometimes you may get a system crash, or you may get a black screen, if you get a black screen you may be able to "back out" of the graphics display with [Esc] or [Return] if that is the right "back out" command key. If you are going to use the "Windows" key to pop out of my programs then click on the "task bar" to get back in, only do so while the programs are in the "Text mode" 80x25 screens, which are blue with white text in the programs most of the time. Sometimes bugs in the video BIOS, or the mouse driver can make a system crash, so using another computer with a different mouse and or graphics card may work better. Sometimes pressing too many keys on the keyboard may "over interrupt" Windows (tm) or your motherboard or something like that and cause a system freeze, so its best not to keep pressing keys or moving the mouse around while the program is re-drawing the Graphics display etc. Anyway the point being that most program bugs will put up a small error message then the program will exit, if the whole system does a freeze then it is probably something more than a simple logic error in the program's code. Real bugs should repeat in exactly the same way when you re-do the command sequence, except maybe for memory faults, but Windows (tm) should detect memory access faults and halt the program and display an error message in the "blue screen of death" as it is sometimes known, if the whole system freezes under Windows (tm) that is not a normal program code error issue, probably. --- Q AND A ABOUT DANCINEC.EXE (tm) Q1) What does de-Bayer mean? A1) A Bayer sensor is named after an engineer at Kodak (tm) that figured out a way to cover a monochrome sensor with Reddish, Greenish, and Bluish filters in a kind of checker board pattern so that in each 2x2 square there is one Redish pixel, two Greenish pixels, and one Bluish pixel. So there are four ways the filters can be arranged, therefore DANCINEC.EXE (tm) has four "Bayer orders" to select from, if you select the wrong Bayer order, the image will have incorrect colors, like blue faces if red and blue are reversed, so you can try all four orders on a sample RAW file that has obvious colored subjects in it until you find the order that works right with your camera's data. The Bayer order depends on the starting pixel on the sensor, so if you shift the active window around on your sensor you might change the Bayer order (1 to 4 in the camera model setup menu, #3 in the #1 setup menu from the program's main menu). Also if you use the RAW data image mirror (flip) commands those change the Bayer order needed. For the known cameras the correction for the mirror (flip) should be automatic based on the X or Y mirror (flip) being active or not, but if you have a new camera and are doing manual configuration you will need to compensate for the mirror (flip) and starting pixel by selecting the right Bayer order to use when loading your RAW camera data. After the program separates the Red, Green, and Blue images from the Bayer "checkerboard" pattern, the missing pixels in each color are interpolated, since Red and Blue are only 25% of the pixels, 75% of those colors are made from interpolated pixels based on the closest "real" pixel, for the Green pixels half are "real" and the other half interpolated from the closest ones. There are options for the way to interpolate the "missing" pixels since some methods may work better on some image subjects and other methods on other subjects, but its not something you need to change too much once you get a balance that works for your camera and normal subject matter that you shoot. Q2) What is Luma and Chroma? A2) Images have two parts, there is the general brightness of colors called Luma which can be like a black and white photograph and holds much of the sharpness and detail in what you see, the other part is the Chroma which can have all brightness removed and just shows the hue of various colors. The eye is less sensitive to detail in the chroma part of images, so the chroma can be softened and otherwise filtered more than the Luma to reduce noise in the processed images. In my finish modes the image processing of the images is split between the Luma and Chroma in each image, and there are adjustments to improve those two parts separately before they are combined to make the RGB result that you will be using. Q3) What is WBC? A3) WBC or White Balance Correction is the basic White Balance KCC file correction used to convert the dark greenish RAW data into a "usable" image. WBC is used alone for some process methods such as workprint to get a rough image you can use for editing. WBC also has curves and so can correct for EI/ISO while it does the white balance and monitor gamma corrections. Finish modes have two additional color corrections, the EDR to improve the highlight detail, and SCP to "touch up" the image color balance, gamma, and curves before the final finished image is saved to avoid needing a second pass of the frames through the program and so saves time doing the de-Bayer and final Grade in one pass through the program. Q4) What is EDR? A4) EDR or Extended Dynamic Range is a second form of Luma that can be used in processing Bayer images, normally the sensor data's Red and Blue pixels data must be clipped to match the maximum subject brightness captured by the Green pixels. Doing that clipping of the red and blue gets you a "white balanced" image without colored highlights and colors that look bright and correct. But there is some extra Luma information in the Red and Blue pixels data above the clip point needed for white balance (WBC). The EDR or Extended Dynamic Range processing blends the color corrected image with the monochrome EDR image (taken from the same RAW data) to improve the highlight detail and avoid the "blown-out" cheep video camera look to the processed images. Because the extra Luma information added by the EDR overlay does not have the same white balance as the white balanced image (WBC) there is some shift in the lightness or darkness of colors in the highlights, but less so in the mid-tones and shadows unless you adjust the EDR KCC in some odd way. EDR does not shift the Hue of tones much, mostly their brightness making red and blue subjects a bit more bright perhaps, but with tonal detail. It tends to make the images look more "film like" because of the extended dynamic range above the green clip point of the sensor. The Red, Green, and Blue sensors are not equally sensitive, the Green are about a half to a full stop faster, so you get maybe a half to a whole stop improvement in some detail in the highlights, along with some color shifts that are not so different from film getting a bit de-saturated in the over exposed parts of the image. You can adjust the EDR blend over a range and not use it if you don't want it, although it can help with well to a bit over exposed images or high contrast subject matter. The color saturation can be automatically corrected in inverse ratio to the EDR blend for you, or you can set the color saturation a bit higher or lower than the WBC image (0 EDR blend). Because the EDR image adds noise to the combined result, additional noise filters are located after the EDR to clean up the image a bit before it is saved for your use as a BMP/CIN/DPX/TIF file. Q5) What is SCP? A5) SCP or Second Color Pass is a "touch up" color correction that is used in the "finish" process methods to make additional small adjustments and "tweaks" to the black level, white level, gamma corrections, and curves. Doing the SCP at the same time as the de-Bayer saves time of not having to do a second pass through the program to get final finished BMP/CIN/DPX/TIF frames for use. Q6) What is a "Project Structure"? A6) To have a place for my Cinema programs to find source files and put result files, commands are included in DANCINEC.EXE (tm) and DANCAD87.EXE (tm) to make a "Project Structure" which is a mass of nested folders on one or more drive on your computer. The folders have an order, Project, then Shot, then Image level, then the name of the image file. In the case of the Divided numbered filename type, the image file is in sub-folders in the Ixx image level folder. Here are some example file paths, C:\P0001\S0123\I01\00000000.RAW E:\P0001\S0456\I01\00000000.DNG G:\P0001\S2978\I01\00000000.TIF H:\P0001\S9781\I04\TIF0001C\234.TIF (i.e. frame 1234) D:\P0001\S3456\I11\PIX0046D\728.PIX (i.e. frame 46728) In the "Divided" numbered filename examples, C means that the folder is the first 500 files ending in 000 to 499, and D means the second 500 files per thousand, 500 to 999. Those are used because when a folder holds more than about 511 files the access for the I/O can slow down, so the "Divided" numbered filename type only saves 500 files in each folder, so you can read the frame number easier the other digits count in base 10, so the C and D are used to split each thousand frames into two folders (A and B at the end of the folder name show that the frame numbers are negative, but since shots are supposed to start on frame zero, you should not see A and B used much). Audio tracks are also stored in track folders in each shot folder, and in the Rxxx Reel folders for the tracks after picture lock. The shot folders are absolute numbers you can use to keep track of where the files for any given shot are, but in the Edit list in DANCAD87.EXE (tm) you can move the "logical" shots around, so there are two shot numbers for each shot, the physical folder where the shot's frames are stored, and the logical shot sequential order, you can see both in the Edit list if you need to cross reference them. If you are not using DANCAD87.EXE (tm) to edit with, using a project structure is still convenient for working on frames with DANCINEC.EXE (tm) as it gives you a place to put the files and keep them in some kind of order. If you don't want to use a project structure you can use numbered source and result files in "any" folder by selecting that option in DANCINEC.EXE (tm)'s source and result filename prompts in the execute menu #3 from that program's main menu. Q7) What is an RGB file? A7) An RGB file means Red-Green-Blue, a image file that holds Red, Green, and Blue color separation images, like three monochrome images that when viewed together on a color monitor or projector make a tri-color image you can look at. The currently (April 2011) supported options are: BMP 8 bpp color or monochrome for source, monochrome only for result. BMP 24 bpp 8 bpc color for source and result. CIN 10bpc 30bpp for source and result. DPX 10bpc 30bpp, and 16bpc 48bpp, for source and result. TIF 48bpp for source and result. BMP 8 bpp monochrome is also supported as a True RAW Bayer sensor data file format when camera type 501 is selected from the camera type sub-menu. Q8) What is an "optical"? A8) Digital processing of one or more set of source image frames to make a result image frame set with effects like those made in an optical printer with motion picture film. Opticals are things like fade, lap dissolve, titles overlay, compositing including static or traveling mattes for split screen and such, and chroma key for green or blue screen shots, as well as making color separations and frame rate conversions like skip frame printing or double exposing frames. Process methods for "opticals" are in the process method menu in its RGB process methods sub-menu. You need to de-Bayer your camera's RAW data to color corrected RGB files, then use those with the RGB process methods to end up with "opticals" in the final result file set. For process methods that require more than one source frame set, or produce more than one result frame set, the source and result filename prompts will cycle through the number of needed filenames, so be sure to check that all the prompts are correct or you will get errors because one of the filenames was not right, or you mixed file types where that was not allowed. Q9) What does bpc and bpp mean? A9) bpc means Bits Per Channel, in this case the channel is Red, Green, or Blue. Common 10bpc CIN/DPX files can be 30bpp, but use 32bits with two bits unused per pixel. bpp means Bits per Pixel used for the image data. Q10) What does bits mean? A10) In a sensor the bits are the number of binary signals out of the ADC (Analog to Digital Converter) used with the sensor. Each bit is 2 times as much signal as the previous bit. Since sensor data is "Linear" and the monitor is at "gamma 2.4" with the image files encoded gamma 2.22 for viewing you "lose" about two bits on converting from sensor data to something that looks right on a monitor, that is it takes about 10 bits from the sensor to get about 8 bits on the monitor. 8 bits on the monitor is not quite enough tones, 2^8 = 256, to prevent all tone banding from being seen in low noise images, your eye needs more like 9 to 10 bits at on a gamma 2.4 monitor from a encoded gamma 2.22 image to not see banding. With that in mind, then you can see that a Digital Cinema Camera should use at least a 12bit ADC on its sensor to end up with 9bits 1 bit above the sensor noise floor, and 2 bits under the highlight clip point. Any exposure error would cause the image to show extra noise or clip the highlights too much, even with 12bits. Newer cameras use 14bit or even 16bit ADC on their sensors to capture a wider range of tones so that higher EI/ISO speeds can be processed before banding shows up, or to allow for larger exposure errors before shadow noise or white clip become major problems. The sensor pixels do not have a "bits" value since they are analog, it is through the ADC that the analog signals from the pixel pre-amp are digitized. If the pre-amp is not adjusted so that there is a 1 f/ stop relationship with the ADC counts of 2x each, you would get more or less dynamic range than taking the data bits as the f/ stop range, that is if the pre-amp is adjusted so that 1 f/ stop change makes 1 ADC count, then a 10 bit chip would have about a 9.5 f/ stop range, a 12 bit chip would have about a 11.5 stop range, and a 14 bit chip would have about a 13.5 stop range. It would be very hard to get the full 15.5 f/ stop range as usable data with a 16 bit ADC, but as things improve maybe sensor and pre-amp noise can be reduced more than it is today. If the preamp gain is higher so that 1 f/ stop makes two ADC counts, then a 12bit chip would have 6 f/ stops of dynamic range. In the case of RGB image files, bits are like bpc and bpp but a bit less clear since 16bpc TIF is assumed to be 48bpp TIF, but a 16bit PIX file is 5bpc for Blue and Red, and 6bpc for Green, while being 16bpp as well. Note that EDR processing in the de-Bayer can extend the tonal range in the highlights to some degree, perhaps half to a full bit, that is half to a full f/ stop, maybe. Q11) What is an OLPF? A11) An OLPF, a.k.a. Optical Low Pass Filter, is a special filter used between the lens that makes the image in a digital camera and the sensor. OLPF are used with Bayer sensors to reduce both aliasing and chroma moire by reducing the contrast of fine detail while not reducing the contrast of course detail much. They are not a "diffusion" filter, as the are made of a "sandwich" of glass and crystals that can divide the light into closely spaced parallel beams. The OLPF needs to be selected to work with the pixel spacing used in the sensor it is in front of so that the light is spread about the width of two or three pixels. To compensate for the blur of the OLPF in the Luma, sharpen software or hardware is used. In DANCINEC.EXE (tm) there are sharpen filters in the "finish" modes to compensate for the Luma blur. Its best not to sharpen the Chroma as that could "un-do" some of the benefit of using the OLPF to reduce chroma moire. Because digital sensors are very sensitive to infrared light, a.k.a. IR, and IR cut filter is usually combined with the OLPF layers to reduce IR getting to the sensor. Likewise UV light should be cut by the OLPF filter to reduce ultraviolet Haze in the shots made. Because the OLPF has thickness and normal lenses are not corrected for a sheet of glass between their rear and the image formed the OLPF degrades the sharpness of the image somewhat more at large f/ stops. The degradation of the OLPF in ratio to the image height is more for small sensors than large ones, because the thickness of the OLPF is greater in ratio to the sensor's image area height on small sensors as OLPF are about 1.5 to 2mm thick, and on larger sensors the image height might be 1/5th the OLPF thickness whereas on small sensors the image height might be close to the thickness of the OLPF. Q12) What is a GVN file? A12) GVN stands for Global Variable Name, its a file used by DANCINEC.EXE (tm) to record how you have the program setup. The name of the current GVN file is held in the file named GCN. You can change the GVN name in the GCN file with the command option #1 in the #1 option from the DANCINEC.EXE (tm) main menu. Do not change, overwrite, or erase the GVN or GCN name while DANCINEC.EXE (tm) is running, ever. When running DANCINEC.EXE (tm) from the command line or from a Batch (*.BAT) file you can change the GVN to use by using the new name as a command line parameter after the program name. You can change the name of the GVN name when the program is running with the #1 prompt in the #1 sub-menu from DANCINEC.EXE (tm)'s main menu. Whenever you are going to start a new setup to prepare for batch processing frames, you should change the GVN name in order to keep the previous setup for later use and adjustment as the program has many setup prompts and if you have to "start from scratch" again it will take some time to review all of the program's many menus. The Graphics Test screen has its own setup configuration file, and some other commands have their own configuration files that are independent of the GVN file, but the GVN file would store the names of some of those other setup configuration files. The make BAT file command in the #3 execute sub-menu of the DANCINEC.EXE (tm) main menu lets you save the current GVN name to a standard DOS console *.BAT file. You can append several GVN setups to a single *.BAT file that way, and then later run that *.BAT file from the command prompt in Windows (tm) etc. to execute several setups non-stop for batch processing of large numbers of image frame files, such as might be running 24/7 for several weeks or longer. To execute a GVN file from the command prompt or from a line in a *.BAT batch file you can write the name of the DANCINEC.EXE (tm), DANCINED.EXE (tm), or DANCINEW.EXE (tm) program followed by the name of the *.GVN file previously setup in that same program version, like this: DANCINEC.EXE MYSETUP1.GVN or DANCINED.EXE MYSETUP2.GVN or DANCINEW.EXE MYSETUP3.GVN You can also manually execute the current GVN settings by using the GVN select command option #1 in the program general setup menu #1 from its main menu, then use the execute command option #8 in the execute menu #3 from the program's main menu. The command line parameter option was included so that the automatic *.BAT batch file command in the execute sub-menu could automate batch processing by passing the name of a GVN file to the program, as each shot or range of shots to process might need their own GVN file setup as needed. The GVN file "remembers" many of the prompts in the program, so to setup the GVN fie you just change the name of the current GVN file, make changes at the program's prompts, then use the [T] for Test command in the execute menu to check if you have prompts setup right, then you can append that setup as part of a batch job with the BAT file maker in the execute menu (#3 from the main menu). Each batch job needs a new GVN name, so remember to change the current GVN name with the #1 option in the #1 menu from the programs main menu, otherwise you will alter the settings in the assigned GVN name file before the needed frames are processed. There is not "limit" to how many GVN files you can make, and you can "re-open" them by selecting them as the current GVN file and make changes as needed, such as the source shot folder number, in order to re-use the GVN to process additional similar frames. In that way you can create a "library" of GVN files for different kinds of frame processing that you carry out more than once. There is ALWAYS a GVN file selected as the current file, so be careful that you make a new GVN or clone an existing GVN file before you make changes at the programs prompts, otherwise you will destructively overwrite the previous settings in the current GVN file. For convenience when doing repetitive batch conversion, such as converting *.RAW or *.DNG frame files to TIF/DPX/CIN/BMP or PIX, and you are using numbered filenames in the edit structure that my programs use for the Edit list, you can change the shot numbers for the shot or shots range to process from the command line like this, For single shots, DANCINEC.EXE ISO320T2.GVN 571 For a range of shots in order, DANCINEC.EXE ISO400D1.GVN 572 583 The command line parameters are: 1) The name of the *.GNV file which you setup ahead of time and can reuse as you need, here the *.GVN files are shown with filenames indicating the ISO and K value, D for Daylight or T for Tungsten, but you would make whatever names help you keep track of what you have setup the *.GVN to do, like 1 can stand for workprint, and 2 can stand for Finish quality processing and such. 2) States the shot number in the Edit list structure that you want to override the current value setup in the *.GVN file. This is the source shot folder, and unless there is a result shot offset setup in the *.GVN file also the result shot folder. The Ixx level should be setup in the *.GVN file, such as I01 for RAW or DNG camera data files to de-Bayer, and I11 for PIX "proxy" files for editing with, and I04 for Finish quality TIF or DPX etc. If you do not state command line parameter #2, then the value in the *.GVN file is used, which may overwrite previously made frame files, so you should setup the *.GVN file to use a shot number, like 9999, that will not be used in normal processing to avoid that accidental overwriting issue. 3) If there are both command line parameters #2 and #3 given, then a range of frames in shot folders to process is selected, all folders between and including the first folder in command line parameter #2 and the last folder in command line parameter #3 will be processed. You should have frames pre-loaded into the source Ixx level in that range of shot folders before you start the frame processing. The last shot folder number command line parameter #3 must be equal to or greater than the number of the shot folder for command line parameter #2. The use of command line parameters #2 and #3 is optional, and parameter #3 is not valid without a valid parameter #2. If parameters #2 and #3 are not stated then the current source and result shot folders selected in the *.GVN file will be used, but it can be easier to run the program from the command line rather than using the menu prompts for just changing the *.GVN name and the shot folder numbers to process, and also you can make a *.BAT file manually in a text editor like notepad.exe to process many shots with various *.GNV files as needed. When you use the *.BAT file maker in the execute sub-menu in DANCINEC.EXE (tm) it does not state the first and last shot folders as command line parameters, those values are setup like all the others by using the program's prompts. You can edit the output *.BAT file to add the first and last shot folders as command line parameters to override the shot folders selected in the *.GVN file for re-use of a *.BAT file generated on additional shots to process. You may get error messages if You have setup some values in the *.GVN file that do not relate to the file paths where the frames to process are, so you need to keep track of what drives and project numbers are setup in the *.GVN files you re-use are, or edit such values to bring a copy of the *.GVN file "up-to-date" with what you are trying to do. Q13) What is a LNC file? A13) LNC stands for Lens Numeric Correction. Because the lens used to make the digital frames may have some optical quality issues, you can make an LNC file for each of your lenses, and even each lens at several f/ stops or zoom settings, the save those LNC files for later use. Each time you setup to process frames shot with a particular lens you setup the execution to use the LNC file that corresponds to the lens used at the time of shooting. Be sure to check that the LNC file selected or not corresponds to the shot being processes, since using the wrong LNC file can make the image look worse rather than better. See the LNC setup menu in the #3 execute menu off DANCINEC.EXE (tm)'s main menu? The selected LNC filename is recorded in the GVN file for later processing of the frames. Q14) What is a DGF file? A14) It is a setup file used for the DNG file type writing and sometimes the reading as well. You need to adjust the Tag 50714 (black level) and Tag 50721 (color matrix XYZ to RGB) in order for the DNG results to be usable in programs that use the Adobe (tm) SDK code to process the de-Bayer of DNG files. The DGF file holds the DNG processing setup information you enter in the DFG setup menus, see the prompt marked 5 = 90150 in the DNG section of the process methods sub-menu. The selected DGF filename is recorded in the GVN file for later processing of the frames. You should not expect the de-Bayer results you get from the 512 byte header DNG files made to produce "finish" quality results because de-Bayer based only on that header type probably only does a minimum level of processing, additional Grading, noise filtering, and anti-OLPF sharpen as well as other image processing would probably be required. The DNG output does give you a way to export True RAW Bayer data as DNG frame files for use in CinemaDNG (tm) based workflows, maybe. The point being, that you need to adjust the tags in the DNG header to match the sensor and OLPF+IR/UV cut filter used in your camera as those things are not automatically known to the program so it cannot fill in those details for you, at least until you make a DGF file for your camera yourself. DNG is a open filetype documented by Adobe (tm) for storage of Bayer sensor data from Digital Cameras, i.e. Digital NeGative file type. Q15) What is a FIG file? Q15) In Graphics display setup the Install command makes a FIG file to hold the settings of the Graphics display, like for Execute Test mode, so that the Graphics setup does not change when you make new GVN files. In other words, most changes you make in the menus of DANCINEC.EXE (tm) are recorded in the current GVN file, but the adjustments to the Graphics part of the program are independent and are setup in the FIG file so they remain constant unless you use the special setup option #2 in the graphics part of the program #4 from the main menu. Q16) How do I re-size my images from camera size to end use size? A16) In the execute sub-menu in DANCINEC.EXE (tm) you can select the Re-Size sub-menu and in there are many presets for common re-size, letter boxing, pillar boxing, and filtering for re-size to reducing aliasing. There is also a manual entry menu for re-size tasks that are not listed in the presets, the source and result image areas have an origin and a width and height. The source image area is resampled to match the result image area. You do need to take care the the numbers you enter make sense, since you cannot have the result image outside the size of the result in total by using the offset values. Only some Process Methods support the re-size, those are in the Workprint and Finish de-Bayer settings and the primary RGB-to-RGB color correction methods. So you need to resampled all elements as needed before compositing. If you are trying to generate PIX files and get image size or re- size error messages, in addition to the current image re-size values, you need to check to see if you have the PIX file re-size options set right for your target graphics mode the PIX files will play back in, that is you cannot re- size the PIX files to 2K if you have their target graphics resolution set to 640x480 since it would not fit. The PIX re-size options are in the PIX setup sub-menu in the program setup sub-menu option #1 from the main menu of DANCINEC.EXE (tm). Q17) How do I get a DOS command prompt? A17) Use the [Start] button in the lower left of the Windows (tm) desktop screen, then select [Run] and type the word COMMAND into the prompt where you would type the name of the program to run. You will then get a window with a DOS command prompt, you can then change the folder manually with the CD command and so on. If you find that the DOS or Console versions show on the screen in some odd way that does not look right for 80x25 text mode program menus, try using the Windows (tm) key on the keyboard, then use the mouse on the task bar to look at the properties of the window the program is running in and try to make adjustments there, and with the re-size bars on the sides of the window the program is running in, like if you only see the first line of the program screen, you might be able to pull down the bottom of the window so you see the full 25 line area and such. You can probably make a "shortcut" to my programs using the mouse buttons when you click on the program file name *.EXE in the Windows (tm) folder, and put that short cut on the desktop to make it easier to start the programs. An undocumented command is that you can drag a PIX filename onto the DANCAD87.EXE (tm) shortcut to see the file display, useful since only my programs display my PIX filetype, but do not do that while the program is running (in background or a window) because that will cause problems (it may erase the file, why I don't know at this time, that in addition to it making the running copy of the program produce a fatal error as you cannot have two copies of the program running in the same folder at the same time, so it would be best to make a copy of the PIX file and drag that if you want to see what it looks like). Q18) I have a window showing DANCINEC.EXE (tm) open but it does not respond to my pressing keys on the keyboard, what do I do now? A18) If you have another program open and are using it, you need to click on the top of the window with DANCINEC.EXE (tm) in it to turn that window back on (the top will turn from gray to blue or something like that). Then the program should respond to your pressing command keys or typing things in from the Keyboard. This is a Windows (tm) "feature" not something I do/did. Q19) I notice that the shadows in my de-Bayered images look greenish while the bright parts look OK, what's wrong? A19) If you are using tungsten lights and you white balanced on a 90% white or 18% gray card, then there may have been fluorescent lights on in the room you were shooting in. It is not possible to fully correct for "mixed illumination" in shots with Grading. You should have used gel filters and filters on the camera so that both your lighting and the ambient light match. The same goes for mixing any two kinds of lights such as fluorescent and window light. In the future turn off any room lights that do not match your light kit's color balance, or optical filter the lights and camera so the kinds of light used match better. In some cases, you may want to make the difference in mixed illumination more pronounced so that the off color shadows look intentional, rather than just poor craftsmanship. Q20) I was shooting outdoors and stopped my lens down to f/22 but my processed images don't look sharp even when I increase the OLPF compensation, whats wrong? A20) Lenses cannot form a sharp image with small openings of the iris because the cone angle of the light rays is reduced. No amount of sharpening can restore the lost detail if you use very small iris openings on small sensor cameras. In general for small sensor cameras best results will be obtained using f/ stops between 3.5 and 6.3, at wider openings the lens aberrations blur the image details, and at smaller openings the diffraction obliterates very fine details. Because you should not stop the lens down to small openings, you will need a set of ND filters to block the un-needed light from the subject when shooting under daylight or other bright lighting. Q21) I am having trouble holding the highlight detail and keeping the actors faces light enough when I shoot them with bright light from behind them, how do I fix that? A21) You can combine the EDR feature and other ISO settings. Also the sensor's dynamic range is not that great with only 12bits, so you need to use a bounce board reflector, or fill lights, outdoors to reduce the subject contrast range so that you can hold the highlights and see the actor's faces. First adjust the lens iris, camera shutter angle, and ND filter to hold the highlights needed, then add fill lighting as needed on the subject. You may need to relocate the camera and change the shot's blocking to avoid shooting things that cannot be made to look good with the range of post tools you can apply. The camera is not a "magic box" that makes anything you point it at look good, although the de-Bayer settings, Grading, and temporal noise reduction can help improve shots, they are not a cure all for poor craftsmanship in your cinematography. Q22) I live in a developing country and want to do a DI with the lowest cost equipment I can, what are the very most minimum things I would need? A22) DANCAD87.EXE (tm) might run on a 80486 133MHz computer with 16MB of memory if you use FreeDOS (tm) as the OS. I say FreeDOS (tm) because it supports large disk drives like PATA 500GB, where as Windows 3.11 (tm) W95 OSR2 (tm), and W98SE (tm) do not support drives up to 2TB, you need at a minimum FreeDOS (tm) or maybe Windows ME (tm) rescue boot floppy (tm) to access up to 2TB drives. DANCINED.EXE (tm) will not run on such a system because it needs more system memory (DRAM), probably at least 128MB to 512MB to run under Windows ME (tm) full GUI install as the OS and process 2K to 4K image files. Also for Grading one of your computers should support at least SVGA video mode 640x480x15/16bit. You can edit using only CGA, Hercules mono, or VGA, but the display is not very good, and SVGA cards with VESA VBE 1.2, 2, or 3 support are very cheep used, you do not need video drivers if you run the 32bit DOS version, DANCINED.EXE (tm) under Windows ME (tm) full GUI install and the video card supports VESA in its video BIOS (like some GeForce (tm) and SIS video cards) (the Windows (tm) Console version uses DIRECT-X (tm) for the graphics display so that needs to be enabled on your computer to use the graphics display). How many disk drives you would need depends on the resolution you want to work at in your DI, 2K files are 1/4 the size of 4K files. If you work at 1280x720 then your storage and speed needs are much less than at 2K or 4K. 1280x720 is about the minimum image size that can transfer to 35mm film for projection without the results looking abnormally soft focus. When shooting its best to shoot 2K or larger for a Bayer camera, even if you are going to finish for a DIY film recorder at 1280x720. 1600x1200 is better for a DIY film recorder since you can output both 'Scope and flat if your 35mm camera in your DIY file recorder has a full aperture gate. If you don't have a 35mm movie camera you can convert an old 35mm projector into a film recorder camera, and expose the color print stock using the projector lens pointed at the LCD monitor using DANCINEL.EXE (tm) to load the edited frames. DANCINEL.EXE (tm) may also run on a 80486 133MHz with 16MB memory, but needs a SVGA video card with VESA VBE 1.2, 2.0, 3.0 and DIRECT-X (tm), to run under Windows ME (tm) full GUI install. If you want to also scan 35mm film into your DI, DANCINES.EXE (tm) can be used to advance the projector, and you can use a DSLR to shoot the frames, DANCINES.EXE (tm) outputs the shoot signal from the computer's parallel port and you wire some small relays to trip the camera's single frame "cable release" input, the software that comes with the camera lets you save the files to a computer over USB as the are shot (but you need a later Windows (tm) computer to run such software, if you record sets of frames to a CF card etc., you can go to an internet cafe and transfer the frames RAW data to DVD disks. To use DANCINEC.EXE (tm) for de-Bayer, color correction, and opticals you would need several P4 class computers of 2GHz or better with 512MB to 2GB of DRAM memory and several 2TB HDD to store and process the frames on. How many computers you will need depends on the image size you are working with as doing a 4K DI takes about four times as long as doing one at 2K on the same number of computers, so if you can work with 10 computers at 2K you might want 40 computers to work with 4K to get the work done in about the same length of time. The batch processing computers do not need their own monitor, mouse, and keyboard you can use a switch box to share those with several computers. Blade computers used for internet tasks may be something you can put into a rack cabinet to take less room than having a bunch of "junk" PC on shelves or tables. You can get small PCB that convert computers with only PATA HDD interface to work with SATA HDD, those seem to work if the drive is the Master and only drive on the cable but there is a jumper that may also let the SATA drive be Slave. Search the third party freeware program fat32format.exe to overcome the intentional crippling of the format command in Windows XP Home (tm) so you can format 2TB HDD as FAT32. Some other FAT32 format programs may allow up to 4TB per disk (I have read that 8TB may be possible but do not have a formatter that does that at the time I am writing this, the 4TB one came with an external USB HDD docking made by UNITEK (tm)). Q23) What is a DI? A23) DI stands for Digital Intermediate. The Digital part is that you have a Digital image frame for each frame in the movie you are making, also called "frame based editing" rather than "linear editing" like used with video tape. the Intermediate part is that the first DI were made to act as a middle step between a film negative and a film release print that would be shown, mostly for VFX work. Today DI stands for the same process of converting the whole movie to image frame files, but the film negative scanning to digital images is replaced with the use of a True RAW recording Digital Cinema Camera doing the job of the scanner but without the film step and much faster, and the film recorder outputting the digital images onto film stock is being replaced with direct digital projection of the Intermediate frame images using a DLP projector conforming to the DCI (Digital Cinema Initiative) projection standard (2048x1080 for 2K source image frames cropped as needed for 2.39:1 and 1.85:1 or 16:9 source images). The file set that get projected in a Digital movie theatre is called a DCP or Digital Cinema Package. A typical move is about four reels or about 88 minutes long, so at 24fps you get 24*60*88=126720 frames in a motion picture, 2K frames are 2048x1556 (or 2048x1536) and at 48bpp you need 6 bytes per pixel, so you get a size of 126720*2048*1556*6=2.422TB, but that is the edited length after the shots are trimmed, to hold the untrimmed frames you may have a ratio of 10x that plus the PIX playback frames and all the audio tracks so you can end up with 20 to 40 TB of files total for working on a feature film, not counting making three backup copies of each file you are working on for protection. Q24) What does NLE mean? A24) Non-Linear Editing, an editing system that lets you access any frame in a whole movie without having to reel through rolls of movie film or video tape. My NLE in DANCAD87.EXE (tm) lets you access any of the frames in any order if your hard drive is large enough to store the PIX playback frames for "random access" viewing. Because the 2K and 4K image files used for "finish" work are very large, you cannot fit all of them on one 2TB drive, but if you have many drive controllers you should be able to hold one Reel of your movie at full size if your computer is arranged to hold that much data, so you can work with 22 minutes of shots at a time to output the edited frames for each "reel" or your project. Since the PIX playback frames are smaller than 2K you should be able to hold all of those on one computer at the same time to view your whole movie project while you are editing it. Q25) What does "process method" mean? A25) DANCINEC.EXE (tm) is made to process many image frames automatically based on the setup you make in the GVN files you make (which can be called automatically from a BAT file, there is a command to make such a BAT file (part of DOS and WINDOWS standard) in the Execute sub-menu). Only one process method is supported per processing "batch" job now. You pick the process method in the process method sub-menu #2 from DANCINEC.EXE (tm)'s main menu. There are various process methods for processing both "workprint" quality image frames to give you a fast look at what you have shot, and "finish" quality image frames that you would use to make a 35mm print, DCI projection master, Blu-Ray, or DVD with. Each process method has a code number, you can see the current process method selected with the code number displayed on DANCINEC.EXE (tm)'s main menu. Some process methods are just for camera development or checking camera features, or when you are doing a "custom" white balance on True RAW sensor data (you need to make a un-corrected TIF frame from the RAW data to look at in the Grading to be able to adjust the WBC KCC file to match the light, exposure and camera/sensor/OLPF/IR_cut/UV_cut type used). Although the primary use of DANCINEC.EXE (tm) may be to de-Bayer and color correct image frames from True RAW Digital Cinema Cameras, there are other types of "process methods" for doing TIF/DPX/CIN/BMP to TIF/DPX/CIN/BMP file or RAW to DNG etc. processing. A process method must be selected before you use the commands in the execute sub- menu. You must also select the source and result file types and the camera model type before you execute a process method, if you do not have a True RAW recording Digital Cinema Camera select Generic DNG as the camera model type. Q26) What does "Bad Pixel" mean? A26) "All" sensors have some pixels that do not work right, in general they are stuck on full "white", but since the Bayer sensor is divided into Red, Green, or Blue a "white" pixel looks Red, Green, or Blue after color de-Bayer and color saturation increase. To "remove" "bad pixels" DANCINEC.EXE (tm) has commands that look at a large number of True RAW image frames shot in "total darkness" to see if any of them are not close to the average "black level" which is not zero signal, but closer to 0.04 (of 0.0 to 1.0). If the pixel is way brighter or darker than most of the pixels, it is marked as being "bad". That "bad" data "template" is then used when your shots are read into DANCINEC.EXE (tm) to interpolate over the "bad" pixels with close pixel data so you don't see the "bad" spots in your image. There is a small blur patch over the bad pixel, but those are hard to see for the most part. Q27) What does "FPN" or "Fixed Pattern Noise" mean? A27) In addition to "bad" pixels, on sensors, you have some patterns that show up as non-moving blemishes in the finished image frames, on CMOS sensors you tend to get vertical lines in the sky or dark areas of the image. By shooting a large set of "black" or "dark field" frames in total darkness (lens capped) then processing those in DANCINEC.EXE (tm)'s special commands for "dark field" process methods (see the Engineering sub-menu in the process methods sub-menu, or the shortcut in the Execute sub-menu) you can make a single "average" TIF frame that holds "only" the FPN or Fixed Pattern Noise of the sensor. Some of the FPN may also be due to the camera's circuits and digital noise. The FPN TIF can be used in DANCINEC.EXE (tm) to subtract the FPN from all the True RAW image frames in your shots as part of the de-Bayer processing to reduce the visibility of the FPN in your finished movie project. Q28) What does "moving pattern noise" mean? A28) Moving pattern noise is like FPN, but it moves around from one frame to the next. Some of it may be caused by digital noise getting into the analog parts of the sensor's circuits or small timing errors that get converted to small brightness changes as the pixels are read out of the sensor. Because the moving pattern noise is never the same twice, its hard (impossible) to remove by subtraction like FPN. What seems to work best is to use Temporal Noise Reduction to average out the parts of the shot that are not in motion so you don't see flickering patterns on walls and such that are not moving, and in the parts of the frame where the subject is in motion, the motion of the subject is higher contrast than the moving pattern noise so masks the moving pattern noise to your eye/mind. DANCINEC.EXE (tm) as some Temporal Noise Reduction for "static" camera shots, for moving camera shots you would need some other Temporal Noise Reduction software that has motion estimation like the third party commercial program NeatVideo (tm), or to re-write your shots so that you use static camera and cut between wide and close-up shots more, rather than pan or dolly etc. Temporal Noise Reduction that uses motion estimation may introduce some motion artifacts into processed 24fps footage. 24fps footage has heavy motion blur so its hard to find the edges of moving parts of the image, with higher frame rates the edges of moving parts of the images are not as blurred at a given subject of camera motion speed. Reducing the shutter angle to get sharper edges introduces another problem of "strobe" effect or "picket fencing", so its better to just not move the camera when shooting for later noise reduction with Temporal noise reduction. Q29) What is a "film scan"? A29) A film scan is a digital image of a movie film frame, usually an image of a color negative frame off Super16mm or 35mm movie film. You can make a movie film scanner with DANCINES.EXE (tm), a movie projector, and a DSLR. Today people are using Digital Cinema Cameras rather than shooting movie film and taking digital "photos" of each of the many thousands of film frames to get their movie's frame set to work with as the do their DI using an NLE. So called "raw" film scans need to be "dust busted" and "de-grained" before they can be used. DANCINEC.EXE (tm) may help with the de-Grain somewhat with some of its de-noise filters, but I have not finished the "dust busting" part of my code so its not released yet. Removing dust from color negatives that are not bleach bypass and so have little silver in them can use a 4th exposure of IR light that goes through the image and just records the dust and scratches for the most part, that is why DANCINES.EXE (tm) has an option for doing the fourth IR exposure in addition to the Red, Green, and Blue exposures. The IR exposure will not help with bleach bypassed color negatives, or black and white negatives since the silver will block the IR light. In most cases some manual review of the frames is needed for "dust busting" to be sure that all dust and other blemishes are marked to remove on each frame's blemish removal mask. Blemishes can be removed by taking data from the previous or next frame, or interpolated over with image tones near the blemish. In some cases manual painting out of a blemish on a scan of a bad frame may be needed, that would be done in a digital paint program by a digital artist. Q30) What is a "film recorder" or "film out"? A30) A film recorder is used to transfer digital images to 35mm movie film one frame at a time so that after processing the 35mm movie film can be used as an intermediate to make release prints from, or used directly to project in a movie projector in a movie theatre. Film out is the process of transferring movie frame images used in a NLE to do a DI so that the digital images end up on the 35mm film intermediate for making release prints. There are several ways to transfer digital image frames to 35mm film stocks. Some use a stop motion movie camera pointed at a small monochrome CRT type, like an ultra sharp TV monitor, others use LCD monitors, or Laser beams to draw the images onto the film one dot at a time (very quickly). Exposure in the first film recorders was about 15 seconds per frame, so with a movie having about 126720 frames (88_minutes * 60 seconds_per_minute * 24 frames_per_second = 126720 frames) you divide by 4 to get the minutes 126720/4 (60/15=4) so its 31680 minutes, or divide that by 60 to get hours, 528, and divide that by 24 to get days, so you end up with 22 days to expose the film at 15 seconds per frame! Today there are some film recorders that are as fast as 24 or 25 frames per second, but others still take more than half a second per frame. In your DIY film recorder the best you can probably get would be about 2 seconds per frame using DANCINEL.EXE (tm), in part because of the computer speed, but mostly because its hard to pump enough light through the LCD panel from a monitor to write directly to color print stock because it has very low sensitivity to red light. To get more light through the LCD you need to take the monitor apart, put a fresnel lens behind the LCD panel, and use a high power tungsten light run at about 85 to 95 volts, maybe 2000 watts so you need heat absorbing glass between the light bulb and the fresnel lens, as well as fans on the LCD to keep it from melting due to the heat of the lamp. Today movie theatres are converting to DCI DCP projection with DLP digital projectors, so film prints will not be needed sometime in the not to distant future, although there may be some need for release prints in theatres that cannot afford to convert to DCI digital projection and will need to close when there are no prints to show. If you as a filmmaker have prints of your movies, you may be able to get them shown easier in the final days of film projection just before all the film only theatres close because big studios may stop making film prints all together. Q31) What is a DCP? A31) DCI digital projection uses a special set of files made in a special way for the Digital projector's file server to find them, both the image and audio frames, these files arranged in standard folders is called a DCP. The image frames may be in compressed JPG2000 file type. You can search the internet for instructions on how to make a DCP master file set on some hard drives, you can then I guess make copies of the master set to ship out to movie theatres. At some point hard drives will not be shipped like film prints in cans, the DCP files would be sent over satellite or internet etc. In relation DANCINEC.EXE (tm) you would be making 2048x1080 or a crop of that size image frame for output, probably 10bpc DPX, 16bpc DPX, or TIF, then use other programs to convert your edited output frames to the needed compressed JPG2000 frames for projection. Search the internet for the third party programs named easyDCP Creator (tm), easyDCP Creator+ (tm), and easyDCP Player (tm). Q32) I see too much noise in my processed images when I watch my DVD or Blu-Ray whats wrong? A32) If you use the "finish" process modes for your de-Bayer, and you had the filters adjusted for best results, then you may have the sharpen too high, or you need to use a lower EI/ISO when shooting. Although small sensor cameras may be able to produce good results for DVD release making, Blu-Ray requires much lower noise levels because the compression used for DVD and Blu-Ray (tm) may "amplify" any tiny amounts of noise in the source images into visable blocks and "mosquito noise" that show at a reduced frame rate because the total frame refresh rate of those compressed formats is lower than the video refresh rate, making the compression blocks stay on the screen long enough you can see them. When shooting you should try to use the "best" EI/ISO for your camera, only, and adjust the exposure and lighting to get the right exposure level for that "best" EI/ISO. In general the "best" EI/ISO will not leave a great deal of head room for highlights above 90% white subject, so you will need quite a bit of fill light if you want the lowest noise frames you can make with any given camera. Another point is that Temporal Noise reduction can reduce noise quite a bit, but mostly on parts of the frame that are not moving, so shooting with static camera can yield lower noise images than shooting with moving camera images, and noise is more important when your images are going out to compressed formats than for uncompressed 35mm film prints where film grain was traditional and the grain updates at 24fps. Q33) Why does my DVD output look better than my Blu-Ray (tm) size output? A33) The more you reduce the images the better the signal to noise gets because the pixels are blended (at least in software that does the resample for re-size in the right way, so methods can make noise worse in the reduced images). You can also get less banding in the image tones because when images are reduced the tones blend so new ones between the old ones are made. Higher EI/ISO may look usable on the DVD version but not the Blu-Ray (tm), so if you need Blu-Ray (tm) release, you should use lower contrast lighting and subject matter, higher lighting levels, and lower EI/ISO to reduce noise as much as you can. Q34) Why do I see bad aliasing in my DVD size image that is not there in the Blu-Ray (tm) size images? A34) If the source Blu-Ray (tm) size images are too sharp to start with, when you reduce them they can go past too sharp and show bad aliasing. In DANCINEC.EXE (tm) there is an option to do an automatic "gauss blur" on the large images as part of the re-size down to smaller sizes. If you have that disabled or the source images are too sharp you may not be getting enough blur before re-size. In the Re-Size menu (in the Execute sub-menu) you can set the automatic blur active, or increase its blur radius or strength so that after re-size the smaller frames have the right amount of edge softness to avoid heavy aliasing. Another source of aliasing could be the camera itself, there needs to be a OLPF with the right blur degree in the camera in order for the Blu-Ray size frames to not alias. Even if you back off on the OLPF compensation sharpen, if the camera does not have the right type of OLPF filter to start with, there may be no way to avoid aliasing in the later stages of the workflow. Q35) What is "workflow"? A35) Workflow is the steps the camera's images go through to get from the camera to the projector in the movie theatre. More particular to DI is the set of software programs used for each processing step of the camera's images as the "flow" from one software program and computer to another on their way to a finished movie. Q36) What is a "Digital Cinema Camera"? A36) A "Digital Cinema Camera" shoots "frame based" images rather than compressed image streams that do not include unique image data from each frame in time, that is the camera shoots 24 totally new images per second. Originally it also meant that the frames shot were recorded uncompressed so that all the detail of motion and tone was recorded, like the way 35mm move film records unique image data for each of its 24fps images. Also uncompressed images were made of 35mm film in film scanners, so Digital Cinema Cameras were like film scanners in that only the highest quality uncompressed image file types were used with at least 10bpc or 16bpc. RED (tm) produced some cameras they seem to call "Digital Cinema Camera"s that record lossy compressed frames, that may be in some ways similar to lossy compressed frames used for DCI projection (DCP a.k.a. Digital Cinema Package) which may use JPG2000 like wavelet compression and perhaps a Log curve to remap the tones from 14 or 16 bits down to 12 bits (?) (but not using Luma/Chroma encoding it seems). Compressing both the camera and projection can lead to "double compression" losses, its better if you are going to use compression to only do it once of the end use format, and since compressed projection is going to be unavoidable it could be better not to lossy compress the camera images also, although what you can see of the loss may not affect movie theatre ticket sales. They also call their lossy compressed images "RAW" which has probably added to the dilution of the meaning of what a "Digital Cinema Camera" could be, since if lossy compression is "RAW" then what is not a "Digital Cinema Camera" since it does not seem to depend on the image being an accurate recording of the sensor's ADC data being recorded with maximum faithfulness. A "Digital Cinema Camera" is a digital image replacement for a film type movie camera to be used by Cinematographers, rather than videographers (i.e. not a live broadcast or ENG camera). Q37) What does "True RAW" mean? A37) RED (tm) started calling lossy its compressed image format, REDCODE (tm), from their cameras "RAW" Bayer sensor data, so to help people understand when I am talking about actual 100% Bayer sensor data as it came out of the sensor's ADC, I usually write "True RAW" to distinguish actual recording of all the data bits coming out of the sensor's ADC from any other lossy versions of such data. The term "lossless" might be sometimes used to mean "visually lossless" meaning some test subject from some distance from the screen could not make out the losses, whereas at other times some may state their data are "mathematically lossless" which gets even more confusing when they take into account some tolerance for what would make that mathematical part of any loss visible or not, so some of these terms are not well defined and should be stated just as "lossy compression" to avoid confusion with actual sensor data as it comes out of the sensor's ADC. Q38) What does "lossy compression" mean? A38) Lossy compression is a recording of images that tries to use some "tricks" to hide missing data that is left out in order to reduce the number of bytes required to store the image. How well those "tricks" work for any given frame of a movie can vary depending on the detail in the image, the subject motion, and other factors as the "lossy compression" generally has a "bandwidth budget" for each frame and needs to reduce some image properties in order to meet its target result image file size. The greater the ratio of compression in general the more information gets altered to reduce bandwidth and image file size. How much loss you can see in a side by side viewing of the with and without compression versions depends on the way you view the results, how far away you are from the image as presented, and other factors. Lossy compression is easier to use on finished graded images since the trick to the viewer will be better known, than when using lossy compression on True RAW data where the image tones would undergo wide adjustments in Grading along with ISO and K compensation, that is one problem with using H.264 as a camera recording format, you cannot make large adjustments in Grading without upsetting in part at least the assumptions of the compression about where on the viewers display brightness the camera's sensors tones would fall in the displayed results. Q39) How do I make a DCP a.k.a. Digital Cinema Package? A39) You can search for a program called "easyDCP Creator (tm)" or "easyDCP Creator+ (tm)" and download a copy, it comes for MAC or PC. You can use DANCAD87.EXE (tm) to edit with using TIF and PIX frames, then after you do the mix-down you get WAV for the audio tracks, and you can use the Aggregate command in DANCAD87.EXE (tm) to make an edited set of TIF full size (HD, 2K, 4K etc.) frames. DANCINEC.EXE (tm) can convert that edited frame set into DPX 10bpc or 16bpc with the needed tone adjustments and clip points for loading into "easyDCP Creator (tm)" or "easyDCP Creator+ (tm)". You may need to split the stereo WAV from DANCAD87.EXE (tm) and convert them from 16bit to 24bit mono WAV for loading into "easyDCP Creator (tm)" or "easyDCP Creator+ (tm)", you may be able to do that in a low cost Audio editing program like Magix (tm). A DCP or Digital Cinema Package is what is used for Digital movie projection in movie theatres in place of 35mm film prints now, the maximum resolution is 2K or 2048x1080 and 4K or 4096x2160, within those the image area is smaller for 2.39:1 or 1.85:1 movie formats. In may not be unknown use the image area is 1920x1080 pillar boxed in the 2048x1080 projection area, with quite a bit of over-projection in the theatre. It may also be not unknown to project both 2.39:1 and 1.85:1 movies using the same lens on the projector and the same curtain adjustments around the screen so that the actual aspect ratio of the source images is ignored by the theatre. You may be able to check your DCP by using easyDCP Player (tm) to make sure it is working before you send it to a movie theatre for projection. Q40) How do I make a 35mm release print? A40) There are two major ways, you can use DANCINEL.EXE (tm) and build a DIY film recorder, or you can send your edited frames and mixed WAV tracks to a Movie Lab for transfer to film at high costs. In both cases you need a set of edited frames and WAV files from the audio mix-down, so you will need to make those for any finished results, DCP, Blu-Ray, DVD, 35mm film print. The gamma of the finished frames and amount of noise reduction needed may vary depending on the end use format, so you may need to make more than one finished frame set at one or more target frame sizes to meet all of the end use formats needs. When you Grade the frames using DANCINEC.EXE (tm) try not to push the contrast too high because you will probably lose some highlight and shadow detail going from your "finished" frames to the end use format or projection, so you want to keep a little "extra" in your graded images, that you know will get lost later when viewed. For 35mm film use your "finished" frames may need the optical sound track offset (images off center and close to one side of the frame) and have mid-tone at the right D-LAD value of 445 or 470 (about 0.462 full scale in monitor images, but re-mapped for "LOG-C" a.k.a. Cineon (tm) type CIN or DPX files) so LAD-tone will be about density 1.03 in the release print (if you are off the LAD target density for mid-tone (LAD tone reference patch is a little darker gray at 16% (0.16^(1/2.22)=0.438) than the standard 18% Gray card (0.18^(1/2.22)=0.462)) then your prints will be too dark or light when projected). As time goes on all projection will move to DCP so 35mm prints will not longer be made or be projected, how many years from the time this is written that will take is unknown at this time, but if things continue as they have been going in about 5 to 10 years perhaps in North America and maybe less than 20 years worldwide. Q41) When I use the WBC KCC for "Workprint" mode the image looks about right, but when I use "Finish" mode it looks dark, what's wrong? A41) If the EDR KCC is set to DEFAULT.KCC and you have the EDR enabled, then the WBC image is being mixed with the un-corrected RAW image which is dark, so you get a dark result, so you need to make a EDR KCC with DANCAD87.EXE (tm), select a valid EDR KCC, or disable the EDR part of the "Finish" processing method. Q42) How do I get the "look" of the ITU/CCIR 601 limits when I make a CIN/DPX/TIF result file? A42) The built in ITU/CCIR 601 limits LUT is just for BMP frames because it applies to 8bit data. To simulate that sort of result you can try using the "Finish" de-Bayer or RGB-to-RGB process method and when you make the SCP KCC file use "Curves#2" and adjust the black point to be above full black and the white point to be under full white, and then adjust the curve points between to make a straight line between those adjusted black and white levels or make a curve between them that puts mid-tone where you want it. The levels would be 16/255=0.06274 for the "black" level, and 235/255=0.9215 for the "white" level. Q43) I made a WBC KCC but when I process the frames the color is all messed up, what's wrong? A43) There may be two things that you could have done that would cause that. The first is that you selected Levels#2 by accident when you were using the Grading commands in DANCAD87.EXE (tm) rather than Levels#1 (or Curves#2 in place of Curves#1). Another thing that is easy to do that can cause problems is that you accidentally processed the "reference" TIF file with a WBC KCC loaded and the wrong process method so that the "reference" TIF file is not an image of the True RAW sensor data, if you do that you make a WBC KCC to correct for that TIF file and not the sensor data so when the sensor data is used to make the corrected frames the WBC KCC has the wrong values in it. You should use the process method 20004 to make the "ungraded reference" TIF frame for doing the Grading, and right after you make it change the result filename and path so that you don't overwrite it by accident. That mistake can happen if you make the WBC KCC, then check the results without changing the result filename and path overwriting the reference TIF with a 1st correction, then go back into the Grading to make additional adjustments and see a very messed up result, because you are looking at the image processed TWICE through the WBC KCC rather than just once as it should be. Those are easy mistakes to make and can be very confusing the first time you make them, so always change the result name after you make the ungraded reference TIF to grade with so you don't accidentally overwrite it with a graded TIF. And always check to see if you are entering Levels#1 or Levels#2 or Curves#1 or Curves#2 when you are grading since mixing those up will also lead to confusion later. Q44) I made some DNG, RAW or BIN result files and the graphics preview looks wrong, like dark and greenish, what's wrong? A44) That is what RAW Bayer data looks like, so sometimes the results of some process methods are shown like that because that is what that process method makes, it would be up to the downstream program in the workflow to correct the color in such result files. Also the Graphics display does not always have the LUT to correct the images from what were produced since that would be another level of processing. If you make dark greenish True RAW type data files, you can use another pass of processing on those RESULT files to re-process them with a process method that does color correction if you want to view them with color correction, normally the Graphics display shows the result files as they were made for review, except for the DPX/CIN Log-C to monitor compensation and the CCIR/ITU 601 limits compensations, which are "built in" to re-process the display of the result files for ease of use in not having to manually re- process those non monitor viewable result types. Q45) I made a DPX or CIN file and the downstream program in my workflow does not show the image the way it looks in your graphics Test display, why? A45) You need to adjust the DPX/CIN output setup so that the clip points are in the right place and the gamma adjustments are right. If you have "inflated" the Cineon (tm) levels of 95 to 685 for black to white when you loaded the DPX/CIN (if that was your source) to 0.0 to 1.0 in DANCINEC.EXE (tm) then you need to set the DPX/CIN output clip points to 95 and 685 maybe, and you also need to set the output KCC and gamma to 1.7 maybe in order to convert the monitor gamma used in the Graphics Test to Cineon (tm) Log values in the result DPX/CIN frames that your downstream program is reading. You can use the "tag" process method that adds gray step patches to a sample frame so you can measure how the image tones change and end up through various programs in your workflow, there is a probe in the Graphics Test (press [P] to get it) that gives you a readout of the image tones you see, you can use those probe values read to help you adjust any LUT or other settings like the clip points and gamma adjustments in any of the programs. DPX/CIN files are not at monitor gamma and cannot look "normal" without undergoing some conversion from Cineon (tm) Log to monitor Gamma, so you need to know what kind of DPX/CIN file you have or are making and what kind the downstream program is setup to expect in order to have things look right at each step in the workflow. Check that you do not have both the DPX/CIN output adjusted AND the Graphics Test display using its DPX/CIN load LUT at the same time since you don't want to compensate for the LOG-C or anti-LOG-C twice since that will overcompensate and give very off results in the downstream parts of your workflow. Q46) I made a DPX file and it looks very wrong or will not load in my downstream program? A46) DANCINEC.EXE (tm) supports BOTH 10bpc and 16bpc DPX in both bigendian and littleendian or at least it was meant to. Some programs my not read 16bpc DPX files, so that would force you to use 16bpc TIF or 10bpc DPX or CIN. Try 10bpc DPX if you need SMPTE time code and cannot use the 16bpc DPX made. I know of one program that shows the 16bpc DPX with red and blue reversed, that seems to be a problem with that program and not my files. If you have problems with using my DPX files let me know and I will inquire about what files you have that do work. Q47) I am having problems loading your DPX files into another program that works fine with my other DPX files, how do I fix this problem? A47) There is a CIN and DPX header trace in the CIN and DPX setup option #6 in the setup menu #1 in DANCINEC.EXE (tm). You should make a trace of your file that works, then make a trace of the file you made with my program, open both of them in two copies of NOTEPAD.EXE (tm) and look to see where the fields are not the same, then go into the DPX header configuration and change the settings so that the files made match your working files better, you can set the result to default and enter a value, or set the option for the field to NA which will output the "all ones" or whatever the CIN or DPX spec says is the equivalent that flags that that field is not being used so the reader should ignore the value in it. If you are still stuck you can try to email me and I will probably ask for a sample CIN or DPX file that works for you to see what the problem might be by looking at the header values myself. The header setup values are saved in the *.GVN configuration file so I might be able to email you an edited version of the *.GVN file you are using along with your *.CIN or *.DPX so you can see what field settings work with your other programs. In addition to the header values are the clip points and gamma or curve adjustments, the filenames for the input and output KCC files are in option #3 in the execute menu #3 in DANCINEC.EXE (tm) so check that you do not have the wrong *.KCC file entered there if your image tones look wrong etc. Q48) How do I tell if I have my film scanner or Digital Cinema Camera *.KCC files adjusted right so that the CIN or DPX files will be compatible with standard workflows? Q48) You should shoot some film of a gray card at normal EI/ISO, and then plus one f/ stop and minus one f/ stop (to start with). You should then measure the film density and if the color negative is processed to "gamma 0.6" the density of the magenta layer should be +0.18 density (0.6*0.30 density per stop) for the plus one stop exposure and the minus one stop exposure should be -0.18 density. In a Digital Cinema Camera where the analog pre-amps are adjusted for minimum gain to get maximum dynamic range the code value should double for the +1 stop exposure and be half for the -1 stop exposure when you are using the probe in DANCINEC.EXE (tm) graphics preview or the probe in the Levels#1 part of the Grading command in DANCAD87.EXE (tm). You need to subtract the black level offset to find the change in code values for the Digital Cinema Camera, in other words you might measure something like this, black offset 80, 18% gray card exposure 480, plus one f/ stop 880, minus one f/ stop 280. Once you get all that right, then you adjust the *.KCC file in DANCAD87.EXE (tm)'s Grading command to bring the 18% Gray card to signal 0.462 (of 0.0 to 1.0 range). Then adjust the curves to get a smooth rounded S-curve. Make a CIN or DPX file in DANCINEC.EXE (tm) using the *.KCC file for the color correction of the "RAW" data from your scanner's camera or the Digital Cinema Camera. If you select the Units in the DANCINEC.EXE (tm) graphics probe for 10 bits, the one stop minus exposure should be about 90 points under the about 470 code for the 18% gray at EI/ISO exposure. The +1 f/ stop exposure will be a little less than 90 points above 470 because you will be using the S-curve to do "soft clip", but without the soft clip both the plus and minus one stop exposures should be about 90 code values from the 470 value for the 18% Gray card. To get the 18% Gray card to read right you need your *.KCC file to bring in the data and correct it so that the 18% Gray card is at 0.462 signal level (1/2.22 gamma correction). To get the 0.462 signal level to 10bpc code value 470 in the CIN or DPX file you need to set the output clip points to 95 and 685 and the gamma adjustment to 1.7, those settings will place black, mid-tone, and white at the right code levels. DPX and CIN files output would normally already be "soft clipped" to have ALL tone values fall between 95 and 685 (or 64 times those for the 16bpc DPX). If for some reason you are not trying to make graded and soft clipped CIN or DPX, you would need to adjust your *.KCC and the output settings to have the clip points at 0 and 1023 and the gamma adjustment to 1.0, and do all your corrections with the *.KCC files, but those adjustments depend on your scanner's or camera's RAW data, you target is to have the base density or sensor black level just above 95 in the output CIN or DPX file and to have the code values increase about 90 points per f/ stop. Doing that may or may not put 18% gray near 470 code value, it depends on the EI/ISO and exposure you are using in the camera and processing or sensor gain variations. Because the film has a S-curve you will not get 90 points per f/ stop near the ends of the range. Because Digital Cinema Cameras double their code value about every stop, the *.KCC file used to convert camera data to Cineon (tm) Log-C needs to increase the gain in the dark areas and to decrease the gain in the bright areas since in a 12 bit sensor you might get 25 points per f/ stop in the shadow areas (after black level is subtracted) and 500 points per f/ stop in the highlights (after black level is subtracted), where as the goal in the Cineon (tm) Log-C file is to have 90 points for ALL f/ stops both in the shadows and highlights if you do not want "soft clip" applied, so the code values will be "overrange" and above white clip at 685. When "soft clip" is applied the f/ stops above mid-tone value of 470 in the CIN and DPX file are reduced from 90 code points to smaller and smaller values to keep all f/ stops above mid tone within the range of 470 to 685 so plus one stop might be 80 points, plus two stops 55 points, plus three stops 30 points and so on until the maximum f/ stop gain is just one or two point per stop at code 684 to 685. Going down Kodak (tm) did not make a lower end soft clip because film has a natural contrast reduction in the "foot" of its S-curve, if you are processing data from a Digital Cinema Camera, you need to simulate the lower end contrast decrease in the *.KCC file so that the lowest f/ stop is quite a bit less than 90 points per f/ stop of camera exposure change. Only the f/ stop right under mid-tone of 470 would be exactly about 90 points, other stop changes above and below those two points would end up having less than 90 points per change in camera exposure. In your scanner though you do try to keep the conversion constant to 90 points per f/ stop or rather 0.6 times the density change that would be the change for one f/ stop because the film already has an S-curve, in other words if you add or subtract 0.18 density filter in your scanner, you should have the *.KCC file pass that to the output CIN or DPX file as a 90 code value change. You do not use 0.30 filters in your scanner to simulate the one stop changes because the color negative film is processed to have about 0.6 contrast near mid-tone (the contrast is lower in the dark and bright parts of the image) so 0.30 density per f/ stop in front of the camera is converted into 0.18 density change in the light path in the film scanner. Mid tone density in the scanner would be about 470-95=375*0.002=0.75 density above base plus fog density in the color negative (unexposed but processed color negative film of the type being scanned). Full range for the scanner would be about 1023- 95=928*0.002=1.856 density above base plus fog. That corresponds to at least 1.856/0.18=10.3 f/ stops of camera exposure range, but in the scanner the scanner camera ONLY needs to record 1.856/0.30=6.186 stops of dynamic range because the film is processed to a contrast of about 0.6 near mid-tone. Sometimes Digital Cinema Cameras may record their 12 bit (or more) sensor data using a "Log" data format like the Cineon (tm) Log-C so that they can just record 10 bits rather than 12 (or more), but in converting the doubling of the sensor data to same sized code steps (in this case 90 points per stop) for each camera f/ stop of light brightness change, they can save on bandwidth and increase record time without much loss in image quality in the processed images. That was part of why Kodak (tm) came up with the Cineon (tm) Log-C system, because you could store usable film scans in 10 bits that way, although with modern grading methods its better to use 16 bits (16 bpc) since Kodak (tm) did not envision other than doing soft-clip ONCE only, and then just adjusting the data values up and down with addition or subtraction, not the more complex curves used today that can produce histogram gaps in Cineon (tm) 10 bit data do to its "just barely good enough" as-is recording of film images which had grain noise that helped reduce some of the tone banding problems with todays heavy grading and re-grading of Cinema images. Digital Cinema Cameras can make more noiseless images, so its better to use 16 bpc DPX files in order to have as much as 64 times the tonal data to work with in grading before histogram gaps show up in the finished viewing. Q49) How do I expose my Digital Cinema Camera in direct sunlight and avoid burned out highlights? A49) OptimaCine.com (sm) got a Chinese version of the MacBeth Color Checker and shot some Daylight tests with their OptimaCine 2.5K (tm) camera. These results may give you some "ball park" values for the ND and IR cut filter values you might need. These tests were shot to see if additional IR cut filters would be needed and to get a rough idea of their camera's EI/ISO speed. Shooting at f/4 plus ND 0.90 density (aka ND8) about 1pm in full sunlight the most white patch of the chart read in a normalized signal range of 0.0 to 1.0: Green = 0.3779 @ 1.4ms exposure time (1/714 second) which is about right since you want the green pixels under 0.50 signal for a 90% white card. Green = 0.7677 @ 3.2ms exposure time (1/313 second), this is a little overexposed since the most white patch on his chart is less than bright white, but might be usable with very soft lighting and dark subjects. If you take and exposure of about 1/500 second as the maximum at f/4 plus ND 0.90 density, and adjust for the rule of thumb you get, f/4 plus ND 0.90 at 1/500 f/4 plus ND 0.60 at 1/1000 f/4 plus ND 0.30 @ 1/2000 f/4 plus no ND @ 1/4000 f/5.6 plus no ND @ 1/2000 f/8 plus no ND @ 1/1000 f/11 plus no ND @ 1/500 f/16 plus no ND @ 1/250 The rule of thumb says that the exposure will be between f/16 and f/11 in Hazy direct sunlight when the exposure time is equal to the inverse of the EI/ISO (ASA) number. So if you take EI/ISO 320 being between f/16 and f/11, then this test shows the camera's lowest EI/ISO would be about 250, normal shooting at about 320, and about 640 would be used for higher contrast subject matter. These are VERY approximate values, but you can figure you own numbers from the exposure and ND value used for the two test frames. Since you should not stop the lens down past its diffraction limit to get maximum resolution, you would be shooting at stops between f/3.5 and f/6.3 most of the time, with the peak resolution being about f/4.5 maybe but that depends on the lens quality. Because of the lens needing to be open to get full sensor resolution, you would need much more ND filter than just ND 0.90 density (aka ND8) more like ND 2.1 to ND 2.4 density, to get movie camera like results with 180 degree shutter. Because ND filters can pass some IR light, and additional IR cut filter would be needed to reduce the IR leak through the ND filters, like a Tiffen Hot Mirror filter. You might stack your filters like this (starting at the subject side): 1) Tiffen Hot Mirror (or other good IR cut filter). 2) Glass ND 0.90 filter. 3) Tiffen Hot Mirror (or other good IR cut filter). 4) Glass ND 0.90 filter (or ND 0.30 or ND 0.60 depending on the time of day). 5) Linear Polarizer (about 1 to 1.5 stops light loss maybe you can check yours). That would end up at about 7 to 8 f/ stops light loss. You need to adjust your light meter to match the reduced EI/ISO the filters cause. The reason for putting the ND between the two Tiffen Hot Mirror filters is to reduce the "hall of mirrors effect" as the dichroic filters act like mirrors for some colors. f/11 no ND 1/500 EI/ISO 500 (about native speed) f/11 ND 0.30 1/250 EI/ISO 250 on light meter f/11 ND 0.60 1/125 EI/ISO 125 on light meter f/11 ND 0.90 1/64 EI/ISO 64 on light meter f/11 ND 1.20 1/32 EI/ISO 32 on light meter f/8 ND 1.50 1/32 EI/ISO 16 on light meter f/5.6 ND 1.80 1/32 EI/ISO 8 on light meter f/4 ND 2.10 1/32 EI/ISO 4 on light meter You want to shoot with the shutter angle at 180 degrees outdoors in sunlight most of the time so the exposure time would be 1/48th of a second at 24fps. The light may be plus or minus about half a stop and the subject brightness also impact the exposure used. Anyway that gives you a starting point for shooting mid-day in full sunlight. For shooting indoors you would use an 80A filter to convert the 3200K lights to 5500K for the sensor, that is about a two stop light loss. You can maybe boost the voltage to the tungsten lights 10v makes the light go up about 100K maybe (for 120VAC lamps, 20v maybe for 240VAC lamps), so you can use a less dark 80B or 80C filter. If you need to shoot without a bluish filter under unknown low K lights you will probably need to do a custom color balance, so bring a 18% gray card and 90% white card with you and maybe a MacBeth Color Checker so you can shoot some test frames to make it easier to find the white balance tracking and color saturation matrix values to use. If you get IR light leaking, you will notice de-saturation of the colors, and white balance shifting from greenish on the RAW gamma 1.0 display to more pinkish or a bit orange. Its best to add IR cut when you can, it darkens the red signal a bit (adds some green cast) but is better than getting washed out colors and extra noise and soft images because the lens does not focus IR sharp. You will need to do tests and have the right optical filters and filter adapter rings of a matte box to get good results. You do not want to shoot with small shutter angles since that will make the results look more "video" than "cinema" quality. Q50) When shooting 35mm color negative film I have over exposed about half a stop in the past to reduce the grain and get a more "solid" negative, should I also do that when using a Digital Cinema Camera? A50) Forget everything you know about shooting 35mm color negative film. Sensors hard clip unequal color balance at the top end, they do not have a built in S-curve like film is made to have, or was when there was enough silver in it to work like a filter for grains under the top layer. Most of todays sensors clip the green pixels at about the same ISO based on the 90% White side of the Kodak Gray card, they clip at about ISO 160 to 200 for 5500K lighting. You can rate the ISO a bit lower by using some of the unclipped data from the red and blue, but there can be a color shift in the highlights, see the DRX function in RED's de-Bayer. You cannot shoot normal subjects with the green sensors above 50% signal on the 90% White card in the frame, because you do not get any of the specular reflections that are above 90% matte white, even matteish white subjects like a whit cotton shirt would have some specular reflections above 90% white card lit at 45 degrees. Most of the time you want the green sensors at 25% signal, I think Panavision (tm) shoots as low as about 16% signal for white in their Panalog (tm) curve, you can look that up at their site. For higher contrast range subjects where you want to hold the highlights, you need to shoot at HIGHER ISO, that means in DAYLIGHT you need to shoot at ISO 1600 or HIGHER to get subjects in broken sunlight and not burn out reflections in the direct sunlight areas while holding the shadows as well. Same at night where contrast ranges can be very high, you need the higher ISO to roll off the top part of the S-curve, not to clip the highlights. To shoot in daylight where there is sun and shadow you need ND 2.1 plus IR cut in front of the camera, you cannot stop the lens down past about f/ 8 and get the full 4.5K resolution. The only time you can shoot at about ISO 160 to 200 in on a title stand where the subject is lit with 45 degree lighting so that the subject has not specular reflections. You can use a linear pola filter to control the highlight specular reflections, most of the time you want to dull them as much as you can optically before the light clips the sensor, and that goes for any digital camera its not a special issue for RED cameras. So: Measure the contrast range of the subject that is important with your light meter. Adjust the ISO based on how many stops above 90% white some of the subject matter needed is. Set your RAW histogram based on the white card, overrange green signal, not to be used except for overexposed footage, ISO 80 - 125 (green clipped overrange, DRX used?) 90+% green signal for copy stand and test chart use only (no head room), obie lighting for people ISO 160 - 250 (11+ bits under 90% white) 50% green signal for low contrast range subjects (one stop head room), overcast ISO 320 - 500 (11 bits under 90% white) 25% green signal for normal subjects with the specular controlled with a Pola filter, studio lighting (two stop head room) ISO 640 - 1000 (10 bits under 90% white) 12.5% green signal for higher contrast subjects, people with sun from the side (three stops head room) ISO 1280 - 2000 (9 bits under 90% white) 6.25% green signal for very high contrast subjects, such as forest with broken sunlight (four stops head room) ISO 2560 - 4000 (8 bits under 90% white) Past that the noise may be too high for full size use, but if you are going to SD or DVD you can de-noise better and shoot at higher EI/ISO values. Once the signal goes off the top of the ADC it does not come back, DRX or EDR do not replace data that is overrange, it improves the luma because the red and blue are about half a stop less exposed due to the dyes in the Bayer filter being darker and the IR cut filter in the camera cutting the red a bit, and the sensor being less blue sensitive. Because the gamma 0.45 correction plus the "hypergamma" curve used to adjust the mid-tone for the ISO spread the darker tones the bits you end up with in the shadow in relation to histogram gaps is about two less than the bits under 90% white, so the 8 bits at ISO 2560 - 4000 ends up showing histograms more like 6 bits, but there is noise and you normally reduce the images from 4.5K to 1920x1080, so that averages the pixels together (its better in that case to de- noise the HD size so the noise is passed through the down size average) and so you get back about 2 bits with a half size reduction. These facts are the same for all digital cameras, the difference you see is how well the de-noise and oversample are done to get an image out of ADC output that has 90% white 4 to 6 bits above the bottom. Compressed cameras such as those using lossy REDCODE (tm) or H.264 may introduce additional shadow noise problems that show up in underexposed shots so you should avoid underexposure where that is the case by using more fill light. Using log encoding of the tone values rather than linear encoding may be one way such compressed cameras can reduce shadow noise issues (but that may introduce more banding in the highlights if the shot is graded down). Some cameras may have too much noise to be able to underexpose the sensor enough to hold the contrast range of the subject, so you have to re-frame and otherwise control the lighting or time of day or weather to fit the contrast range within the usable range the camera can deliver in the end use size, what looks too nosy for Blu-Ray by be able to be de-noised for DVD, so you need to learn what the limits ARE for any given camera before you find out the hard way that what you shot is not going to be usable. Optimum exposure is just that, there is no exposure latitude for the best exposure for any given camera used on any given subject matter, you need to adjust the lighting, bounce board, and other factors having to do with the subject such as the time of day that you are shooting as well as the camera settings and post production tools to work together to get great looking images for your final grade into your release. Q51) Why is the dynamic range of the sensor less than the number of ADC bits? A51) Because the sensor has "linear" output such that you get "twice" the data for each additional f/ stop, the black level the sensor is adjusted to not be zero ADC counts, but about 0.04 signal level, or 0.04*4095=163. That's why you don't get 12 stops of exposure range (with 12 ADC bits), but only about 11.5 stops because black is not set to zero ADC count. When you raise the voltage on the input of the ADC so the black level is at 0.04, you push off some of the highlight detail from the ADC maximum count, unless you can reduce the pre-amp gain to be less that 1x. Q52) I noticed my highlights are pink, how do I fix that? A52) It seems that the pre-amps or ADC or both are somewhat non-linear near their maximum brightness range. Because the green signal is NOT clipped its non-linear top end is used in the final image. Because the red and blue signals are clipped much lower for white balance, their adjusted "top" end then becomes linear more or less, while the green top end is not because its full range is used. The way to fix that is to also make the red and blue roll off a bit near their top end, or to stretch the green a bit at the top to re- linearize it. Because you want top end roll off to simulate the rounded shoulder of the film's S-curve anyway, its better to roll off the red and blue a bit more than the green with the Curves#1 in DANCAD87.EXE (tm)'s Grading command. In other words, the highlights seem pink because there is not as much green brightness in them, so you need to bring down the red and blue brightness at the top end to simulate the sensor, pre-amp and ADC non-linear curve near signal maximum. Most analog circuits have reduced gain near signal maximum limits. The adjustment needed varies with the color of the lighting used. Some non-linear flare in the micro-lens array on the sensor and such may not be able to be corrected because it does not vary equally for each color since the red, green, and blue sensors have un-equal sensitivity at their actual top clip points, so that can make a colored ring around highlights based on the color of the pixels that are most sensitive, normally the blue pixels when shooting in daylight, so you get a bluish or pink halo around clipped highlights sometimes, but the "core" can be made closer to neutral balance as noted above. Q53) Will HDR solve all my exposure and lighting issues and let me shoot anywhere without using lighting? A53) HDR can be obtained by using two cameras that do not have the "black sun" problem with their sensors in a mirror rig like used for 3D shooting except that there is no spacing of the lenses. With HDR one camera shoots the mid- tone to shadow information, and the other camera shoots the above mid-tone information for the highlights. The exposure "spread" of maybe 2 to 6 stops can be achieved by using ND filters plus IR cut on one of the two cameras, or the shutter angle can be reduced for the highlight camera, although its better to have the shutter angle the same to reduce motion artifacts. The two camera's shutters must "roll" in the same direction and be in sync within maybe +/- 1ms on their start of each frame. Some cameras can take two exposures of each frame, in effect running 48fps for 24fps use with staggered shutter angles, like 5 degrees, 180 degrees, 5 degrees, 180 degrees and so on. It seems too often when shooting you end up with too much or too little light when shooting in natural or location settings. That's one major reason that big filmmakers like Stanley Kubrick shot The Shining (1980) and Full Metal Jacket (1987) "exteriors" in the studio, so the light and lighting ratios could be controlled. Part of the beauty of studio films over location shooting is their having less wind and more consistent lighting. Because the sensor gain is always the same, the ONLY thing that changes from shot to shot and frame to frame is the light falling on the sensor. So you as the DP need to know what the sensor limits are for the camera you are using. In HDR shooting you have TWO exposure levels both have their own clipping and noise issues. To avoid seeing the clipping and noise in each of the two HDR images you use an S-curve on both images that favors the center of each of the two exposures for each frame so that the two images overlap smoothly and do not show heavy noise that would mix with the other images's tones. HDR modes will help people avoid serious exposure flaws, more than just using DRX or EDR which may use the unclipped red and blue sensor data to reinforce the missing green signal above green clip, but to get best results one would still want to center the subject's reflected light on the sensor so that the two ADC counts are large enough to avoid banding and noise, and still be low enough to give enough highlight range above 90% white card ADC output level. With Digital Cinema Camera RAW data it is "easy" to get any EI/ISO over a wide range from the gross image tone standpoint like having the 18% gray card at 0.462 signal level in the graded images (normalized in scale 0 to 1). You cannot adjust film scans over such a wide range because they have much higher fixed pattern noise and grain. But being so easy to later "fix" poor exposure done in the camera does not make up for getting the exposure right at the time of shooting by using the right filters and lighting kit. Its wrong to think that HDR will make "fantastic" images without using lighting, lighting is used to make things look better on the finished display. Because monitors and digital projectors have a contrast range lower than what the camera and eye can see, you still need to reduce the contrast range of the subject matter to look good on normal display devices, you do that in grading where you decrease the camera's brightness range of about 11.5 stops, or more in the case of HDR, down to 5 or 6 stops that the display can reproduce in real world viewing conditions. Grading and lighting at the time of shooting need to work together in that contrast range reduction to get the "best" results on the movie screen or monitor at home... You still need to use an "eye light" and some fill or bounce board even with HDR being used to be cinematic on the final grade. The main advantage of using HDR is to reduce shadow noise and improve extreme highlight detail by shooting with "split" exposures that are optimized for each end of the range needed. Fusing two images from a mirror rig may reduce the resolution somewhat because of the need to re-size, crop, or rotate the images before they can be aligned for fusion. HDR is a tool for shooting shots that need very high contrast subject matter, but even so, the results may look better on the screen if you have enough fill light so you can see the actors eyes and such. Q54) How do I correct for red and blue reversal? A54) You should be able to reverse the red and blue chroma matrix values in the KCC file used for color correction, or make a special KCC file to do that reversal and add a second RGB to RGB color correction pass. The values would then be: Red output Red part = 0.0 Red output Green part = 0.0 Red output Blue part = 1.0 Green output Red part = 0.0 Green output Green part = 1.0 Green output Blue part = 0.0 Blue output Red part = 1.0 Blue output Green part = 0.0 Blue output Blue part = 0.0 Q55) How do I use more than 512MB of system DRAM in my PC running under Windows ME (tm)? A55) There seem to be two ways to do that, the first one is free, with only 512MB of RAM in the computer, or editing the harddrives files with NOTEPAD.EXE (tm) from another computer find the file named SYSTEM.INI in your C:\WINDOWS folder (or the folder your PC uses if you did not install ME (tm) in the default file paths). Next make a backup copy of the SYSTEM.INI file in a new folder called anything you like. Then open SYSTEM.INI file in the WINDOWS folder and edit these parts so they look this way, you need to add the strings shown under the label that you should see in the file, do not delete any of the strings you see just add new lines for the new strings: [386Enh] MaxPhysPage=40000 [vcache] MinFileCache=0 MaxFileCache=524288 then save the changes, quit Windows to shut down the computer, and swap the 512MB memory for a 1GB memory, or add a second 512MB memory etc. to get a total of 1GB memory in the Motherboard's slots. That should let the computer re-boot with 1GB installed. To use a maximum of 4GB under ME (tm) (more like 2GB with my current programs, or as your computer's Motherboard limits, perhaps to 3GB) there is an about $20 program you can purchase that claims to patch ME (tm) to use more memory. It is called "Windows 95/98/SE/ME RAM LIMITATION PATCH v7.0" made by Roudolph R. Loew, he has a 10 minute "demo" version you can try for free to see if it works on your computer, it seems. You can search for his web site for more information, it seems you can paypal him about $21 and he would then email you the full working program file in a zip. You will need to check for yourself as to if this will meet your needs. Q56) How can I run programs that need Windows XP (tm) under Windows ME (tm) so that I can avoid some of the issues with OS higher than ME (tm) with your programs? A56) There are two free programs you can install that claim to do that to some extent, they are called unicows.exe and KernelEx-4.5-Final.exe and work together. If your computer is connected to the internet when you install KernelEx-4.5-Final.exe the installer will dial into someplace and find unicows.exe for you and automatically install it. How well that works you will need to test with the programs you are using. You can search the internet for the needed files. Q57) How do I format large 1TB and 2TB harddrives with FAT32 or FAT32 (LBA) under Windows XP Home (tm) since it limits the size to 32GB? A57) Use the disk manager program in Windows to give the drive a drive letter, but do not use it to format the disk, there is a box you can check that says not to format. Then download the freeware program fat32format.exe and use that or another program to format the new drive letter you made, the quick format only takes a few seconds even for 2TB size disks. I got a FAT32 formatter that says it will format 4TB disks with a USB harddrive adapter I purchased called UNITEK (tm). FAT32 disks can be opened on almost any PC system, MAC, PC, or LINUX, it seems, so having you data on those disks lets you edit the image or sound frames without the need to copy the files from one disk to another. The USB to SATA adapters are convenient to setup and format disks before you install ME (tm) since the disk management tools that are in XP Home (tm) are a bit easier to use. Also you can download free disk partition management tools from the internet that let you adjust the size of disk partitions should you wish to adjust them on your other computers harddrives as you progress through your workflow development. Newer disk clone software you can get free off of harddrive makers web sites lets you copy disks using USB adapters so you don't need to be running ME (tm) to clone a ME (tm) disk for protection, you can do that maintenance using external USB adapters on a XP Home (tm) computer etc. Q58) How do I use a SATA 1TB or 2TB harddrive on older junk computers that use PATA EIDE type disk drives? A58) There are small PC boards that you can get to use a hardware adapters that convert the PATA to SATA so that the SATA drive works on older computers. Very old PC computers will not work with 1TB or 2TB disks, you can only access about 128MB of disk space, so you should not try to use junk computers that are very old, although you may be able to use the case, CD drive, floppy drive, power supply, and upgrade the Motherboard and processor to P4 2.2GHz or faster level so you can run Windows ME (tm) and 2TB or larger harddrives. Used mother boards or even ones a few years old and lower cost now can be had for maybe $40 so the upgrade is not high cost if you have all the other parts from a PII or PIII computer. My programs do not need "drivers" to run, so you can run the computer with minimum hardware and Windows (tm) default if you need to probably, such as if you don't have the video board driver CD and the computer runs in 640x480 mode, my DOS programs can switch the video up using the video board VESA VBE 1.2, 2.0, or 3.0 BI0S to 1920x1440x32bit or 2048x1536x32bit depending on the program and what the video board supports in its BIOS modes. You can always try to search the internet for any needed drivers if you don't have the CD that came with the graphics card etc. The PATA to SATA circuit boards can be found for less than $10, so they are not a high cost item. Some seem to have a jumper on them that may shift between Master and Slave drive or something like that, but you may find some issues trying to run two SATA drives on a single PATA cable, as you may need to change the jumpers and try the order on the two PATA cable connectors. To be safe plan on just using one SATA drive on each PATA controller, since you don't need the CD all the time once ME (tm) in installed on your "render farm" computers you can unplug the PATA cables from the CD and plug them into the PATA to SATA circuit board (its about 3/4 inch by 3.5 inches and does NOT use a PCI slot, it goes on the end of the PATA cable). That way you can use two SATA drives on the render farm computer, one for system boot and another for 2TB of image files for processing etc. If you just have one SATA drive per "render farm" computer, then all your "data" drives also need to be bootable, in that case make two partitions, one about 29GB for the ME (tm) system files and basic programs, and the remainder of the 2TB of disk space for the image files and my program's edit list file folder "structure" (you just need the range of structure folders for the image files needed for that batch processing job, such as if you are processing shot folders 1000 to 1100 you don't need to make folders 0 to 9999 on that disk drive as that would use file space with a large number of empty folders, with my programs "structure" making commands you can expand the number of shots in the "structure" later if you need more folders to put more shots and images or sound into etc.) Q59) Will a polarizer filter increase aliasing and or moire in my digital images? A59) It might, some OLPF filters may use small crystals that can split light into two beams based on the right hand and left hand polarization. If you block all of one or the other kind of polarized light, then you may get asymmetric division of the light passing out of the OLPF. You can shoot some shots and rotate the polarizer filter slowly, then make a zoom crop of several scan lines and see if the aliasing or moire changes as the filter rotates. You can try both linear polarizer filters and circular polarizer filters to see if there is any difference in how the affect aliasing and moire in your images. Q60) How do I get the lowest noise images? A60) The answer is not to overexpose the sensor. The way to get the lowest noise images is to control the contrast of the lighting ratio on the subject at the time you are shooting to get the maximum contrast without exceeding the useful range of the sensor, that way you will not need to increase the contrast in grading that would also increase the noise in the images. Likewise, if you want bright colors in your end results, you should use bright colors in the subject matter you are shooting so that you do not need to increase the color saturation later. Also if you want a particular color balance in your end result it can be best to light for that color balance so as to not to make adjustments in grading. Its common to shoot with neutral color balance then to grade the images to the wanted color balance, but any change requires an increase of contrast to part of the curve of one of the colors and that makes noise worse. Q61) I am shooting with a camera that uses compression or has high noise at low light levels, how do I get good results? A61) First do not use moving camera since that will cause more problems when you apply temporal noise reduction. Second use enough fill light and bounce board that there is some signal in the shadow areas of the image. Third try to rim light or at least have some light on the background so that the whole frame is not dark and you can make out the outline of the subject. Fourth use an eye light so that the actor's eyes are clear and well exposed. Fifth don't time up the image, in fact it can be good to over expose a little then make it darker in grading as any "black lift" will show the shadow noise and compression artifacts more than without any grading up, since there is less noise in the upper midtones its better to overlight and grade down than to under light and grade up, that is why in the past cinematographers shot "day-for-night". Q62) I can't get the full screen graphics preview to work on my notebook computer, what should I do? A62) Use DANGUIVU.EXE (tm) or set the graphics viewer to some program like MSPAINT.EXE. See the graphics config in DANCINED.EXE (tm), DANCINEC.EXE (tm), or DANCINEW.EXE (tm), its option #4 then #2 from their main menus. Q63) I have a quad core processor how do I run more than one copy of your de- Bayer program at the same time? A63) If you want to run more than one copy of my programs at the same time be sure to make several folders and install duplicate copies of the programs so that each has its own folder, you cannot and should not run two copies in the same folder as the temporary files would conflict. I am not sure that you will get much advantage since the computer memory and hard drives are shared, but you may get some advantage. In general you may get more through put using several single core computers than one multi-core computer. When you run my programs its best to close all other programs, and to stop programs running in background so that more CPU cycles are free. Do not try to work on the same shot with two copies of my de-Bayer program as they cannot open the same frame and work on it at the same time, you would need to divide the range of shots being processed so that there is no overlap between the copies of the de-Bayer program's tasks setup. Q64) The program crashed and shows "Run Time Error 216" (or similar) and then windows makes a pop-up that says "Virtual Memory has been Expanded" or words to that effect. Should I report this "bug"? A64) You should check that you have no programs open when running my programs, and in particular DANCINED.EXE (tm), DANCINEC.EXE (tm), or DANCINEW.EXE (tm) as they can use most or all of the computers DRAM. Another source of this problem seems to be that if you open a file folder that has many image files in it and you have the "Thumbnail" view selected, Windows (tm) will use all available memory to make "Thumbnails" of each image file, so when you then try to run my programs there is not enough free DRAM for them to function. If you restart and Windows (tm) tries to run my programs using Virtual memory on the Harddrive then the program my function very poorly and slowly if at all. The answer to this "Thumbnail" issue is to not to use any "Thumbnail" views when you open folders while running my programs as you will probably get low memory problems, and in general not to have any folders open or other programs open or in background while my programs are running, and so in that way insure that the maximum amount of real DRAM is free for their use as needed. Q65) I am trying to set the "Test" process frame, using Execute menu option #7, to a number higher than zero, but all I get after I press [T] for the test in the Execute menu is frame zero, is this a bug? Q65) If you select a "Test" frame that is outside the range to process selected in option #1, Source files for processing, in the Execute menu (#3 from main menu) then the Test frame number is "adjusted" to fit within the range previously selected. To avoid that issue, select the processing range of 0 to 99999999 frames, the program will then process up to that range or stop when no file is found using Execute option #8. If you select a frame number with Execute menu option #7 that is outside the actual range of frame file numbers that are in the source shot folder, then you will get a file not found error, or a no frames processed error, probably. The automatic adjustment of your selection in Execute menu option #7 is to avoid error messages by showing you or processing some frame from the same shot if the one you select is not there. Q66) I selected a guide element and then when I process the frames some other guide element or the same one with different adjustments shows up. Is this another "bug"? A66) If you are using Kinefinity.com (sm) KineRAW (tm) camera models (my code camera model code 504 for the Kinefinity.com (sm) KineRAW (tm) camera series), then guide elements 1 through 6 can be overridden by the meta-data embedded in the DNG source frame files. If you go to the camera model selection menu (#1 then #3 then #5, from main menu) after you select camera model type 504 and the general items, there is a Y/N prompt asking if you want to adjust the meta-data settings, answer Yes to that prompt and you can then turn off (or on) the meta- data override of the first 6 guide elements (such as frame markings and head and tail sync mark flag symbols for the auto-clapper-slate). The guide elements (setup through the [B] for Burn-in option in the Execute sub-menu) numbered 7 through 12 are not overridden by the camera model type 504 meta- data, so you can use one of those to setup some guide element that you want Burned-in to the processed frames all the time, regardless of the meta-data override being enabled or not. Q67) I am using the sub-title dot command files in the shot folders, i.e. TITLE4.TXT, and when I process the frames I do not get the right font file showing, what's wrong? A67) When you make the dot command files as TITLE4.TXT in each shot folder that needs sub-titles, you need to make sure that .f for the font name comes after the dot command that selects the font cell size. The font cell size dot command sets the default font filename for that size so that a valid filename is present, but if you want another font you need to use .f after it to select that font name. The font name selected must match the font cell size selected or the program may show incorrect symbols or report an error reading the font file. If you see blank spaces where the symbols should be it is probably because you selected a font name for a font larger than the cell size selected. Q68) When doing a manual setup for the camera type using camera brand #1 for generic RAW data, there is a prompt that says its for an bit encoding code and to see "documentation", what documentation should I look at? A68) Right here is a good place, for that prompt at the time I am writing this there are these options, all of which relate to 12bit packed pixels modes where there are two pixels in three bytes or four pixels in six bytes, 0=Adobe (tm) 12bit packed two pixels in three bytes which is used in DNG frames, 1=Asia 12 bit packing two pixels in three bytes which is incompatible with the Adobe (tm) bit order and was used in some cameras under development in Asia, 2=Sumix (tm) 12bit packing of 4 pixels in 6 bytes. I noticed something odd with the Sumix (tm) packing when the Bayer order is changed, I don't know if that was from the data being processed by the Sumix (tm) camera software or some bug, I sent them my bit un-packing code and they looked it over and said that it seemed to be the same as the method the use. So I'm not sure what the deal is with the Sumix (tm) unpacking code, as I do not have their packing code I cannot see if the two mate 100%, when the "right" Bayer order is selected the images don't show any strong anomalies, but I am mentioning it as if you see anything odd de-Bayering Sumix (tm) camera data, it might be good to contact me by email so the matter can maybe be looked into more, perhaps. Q69) How do I make my own fonts for use with the Burn-in window-dub and sub- title commands? A69) If you press [B] from the execute sub-menu and then select the setup option menu, there is a command in there that can process the *.814 files into *.F14 or *.FON files. Those file extensions are legacy from the font files in DANCAD3D.COM (tm), they make up three versions of each font. The *.FON files are vector fonts used in making drawings with the CAD programs. The *.814 files are raster graphic fonts that have a fixed cell size, in the initial release of DANCINEC.EXE (tm) et al. there are six sizes, 5x8, 8x8, 8x14, 16x28, 32x56, and 64x112. They are all given the extension *.814 (unlike in the CAD programs where the 8x8 is given *.808). The *.F14 type files are a special conversion of the *.814 files that have the single pixel binary dots enlarged into letters and symbols in text form, so *.F14 is a *.TXT file type. The point of the *.F14 file is that you can edit the automatic conversion of the *.FON file into a *.814 file by using a normal text editor. In addition you can make the right cell size *.F14 file and "draw" pixel by pixel any font you want that can be drawn within the pixel size, the maximum being 64x112 at the time of v0.01 release. You can also use the CAD programs to make a vector font, then use the commands in DANCINEC.EXE (tm) to convert that "automatically" into a *.814 font through vector to raster conversion, but some manual editing of the intermediate *.F14 file can improve the look and legibility of the resulting *.814 font file for use with the Burn-in text guide element option menu commands. If you edit the *.F14 files pixel by pixel for each character you can maybe improve on the automatic vector to raster conversion's results, or alter the characters as you please. If you enlarge the image to burn-in on with the active doubler in the Resize sub-menu, burn-in on the enlarged image then reduce to the final size the results may look better and you can get finer control over the size of the characters. Q70) When I start DANCINEC.EXE (tm) or DANCINED.EXE (tm) from the DOS command prompt and use DANGUIVU.EXE (tm) or some other external program as the graphics preview viewer when I close the viewer I am left on the desktop, how do I avoid that? A70) If you start DANCINEC.EXE (tm) or DANCINED.EXE (tm) from a windows file folder open on the desktop, then the programs can open inside a small window on the desktop, in that way when DANGUIVU.EXE (tm) opens in its Windows (tm) window, it will return you to the DANCINEC.EXE (tm) or DANCINED.EXE (tm) window when you exit it. But if you use DANGUIVU.EXE (tm) from DANCINEC.EXE (tm) or DANCINED.EXE (tm) running full screen mode, and you use a Windows (tm) GUI viewer program Windows (tm) can dump you back at the desktop after the GUI viewer closes, rather than restore the screen to full screen mode as it should, so you need to manually click on the button at the bottom of the desktop or use [Alt]+[Tab] to get back to DANCINEC.EXE (tm) or DANCINED.EXE (tm). If you use the internal graphics viewer that runs full screen in DANCINEC.EXE (tm) or DANCINED.EXE (tm) you do not have the problem of being left on the desktop since the desktop is not active while the program is running unless you use the windows key to pop-out to the desktop. Do not pop-out of my programs to the desktop while my programs are in a graphics display mode as windows may not be able to restore you back into the programs, in general you may be able to pop- out when a blue screen with white text is showing. To avoid having to manually reenter my programs after DANGUIVU.EXE (tm) quits, start the calling program from a Windows (tm) folder by clicking on the *.EXE filename of DANCINED.EXE (tm), DANCINEC.EXE (tm) or DANCINEW.EXE (tm). Only DANCINED.EXE (tm) and DANCINEC.EXE (tm) support both windowed and full screen configuration, DANCINEW.EXE (tm) is GUI and runs in a small window and calls DANGUIVU.EXE (tm) as its default graphics viewer which is also a GUI program and runs in a graphics window. When you use DANGUIVU.EXE (tm) go into the calling program's graphics setup (#4 then #2 from its main menu) and change the DANGUIVU.EXE (tm) window size to be equal to the desktop width and -13 pixels from the desktop height, so if you have a 1920x1200 LCD monitor you would enter X=1920 and Y=(1200-13) or Y=1187. Q71) When I use DANGUIVU.EXE (tm) as the test preview display program from DANCINEC.EXE (tm) the screen "re-draws" twice when the program opens, is that a bug and how can I stop it from doing that as it wastes time and is annoying? A71) DANGUIVU.EXE (tm) gets commands from Windows (tm) to re-draw the window it is in, so if you move the window and such it will be commanded to re-draw, so the second drawing is the result of a Windows (tm) command. The reason that Windows (tm) is sending a "re-draw" command to DANGUIVU.EXE (tm) may be that you have the window active area size set to the desktop width and the desktop height minus 13 pixels, and you also have the "task bar" on the bottom of the desktop set to active and "always on top". For some reason, Windows (tm) first draws the DANGUIVU.EXE (tm) window with the "task bar" over it, then says "Oops" and removes the "task bar" and tells DANGUIVU.EXE (tm) to re-draw over where the "task bar" was. The way to fix this is to de-select the "task bar always on top" option in the Windows setup, you can do it this way: 1) Right Mouse Click on the "Taskbar". 2) Left Mouse Click on "Properties". 3) Left Mouse Click on Taskbar pop-up menu "General" Tab. 4) Left Mouse Click to un-check box "Keep the taskbar on top of other windows" (in XP Home (tm), in ME (tm) it says "Always on top", it may say something else in other versions of Windows (tm)) In that same Taskbar pop-up menu other settings that seem to work are: "Lock the taskbar" set to "un-checked" "Auto-hide the taskbar" set to "un-checked" "Keep the taskbar on top of other windows" set to "un-checked" "Group similar taskbar buttons" set to "checked" "Show Quick Launch" set to "checked" "Show clock" set to "un-checked" "Hide inactive icons" set to "checked" If doing that does not work, let me know what Windows (tm) version you are using and what options are listed for the taskbar or other things that can be set to "always on top" in the desktop. Some programs may force something to be always on top, try closing all other programs when my programs are in use. Q72) The program crashed while I was doing grading, but did not report any errors only the whole computer froze and I could not even get [Ctrl]+[Alt]+[Del] to work, is that a program bug? A72) It seems that if you press the [*] key while in the Grading control screens to re-draw the screen so you can see the grading changes, such as in Curves #1 or #2, and press keys on the keyboard before the screen has re-drawn, and you are working on a computer that has a SATA HDD plugged into a USB harddrive adapter, something sends windows into full freeze that can only be resolved with the hardware re-boot button or manually turning the computer off then back on. It may be an issue in Windows XP home (tm) the computer BIOS, the SATA BIOS, the USB HDD adapter driver or the video BIOS or the graphics card, or I am not sure what else, maybe a combination of those both trying to use the limited DOS memory at the same time. The only resolution I know of at this time is to not to press any keys on the keyboard or roll or click the mouse while the screen is re-drawing after you press the [*] key to re-draw the screen. You can try these work arounds: 1) Transfer your shots that need grading to a EIDE HDD connected to your motherboards HDD controller and unplug any USB drives before you start to grade the shots, then copy them back to the external HDD over the USB later if needed. 2) Try to install a PCI SATA HDD controller card and use that in place of the external USB HDD adapter. I have not tested this option as of the time of writing so I cannot be sure that it will fix the issue. 3) Re-boot your computer with FreeDOS (tm) CD and copy the files to one of the HDD on your motherboard's HDD controller. This may also work with the Windows ME (tm) rescue boot floppy, but all your HDD must be formatted FAT32 or FAT32 LBA etc. in order for FreeDOS (tm) or Windows ME (tm) to be able to read and write to them in the normal way, see freeware programs that let you format HDD up to 4TB as FAT32 or FAT32 LBA (you need to do that before you install Windows XP home (tm) as it was crippled to not format FAT32 or FAT32 LBA on disks larger than 32GB by Microsoft (tm) in order maybe to force people to use NTFS making their disks less mobile to being read and managed under other OS like Linux (tm) and Mac (tm), or Windows ME (tm), maybe). I will be looking into this issue and what other workarounds there might be, but if you have this issue, try not to press keys on the keyboard while the screen is re-drawing, I know its hard to do when working, but it may be enough to avoid the issue. If I am able to port the Grading code to a Windows (tm) program at some point this problem may go away since the graphics would be managed by Windows (tm) rather than the graphics card BIOS. Q73) How can I estimate the EI/ISO of my RAW recording Digital Cinema Camera from the RAW data? A73) Put a pure white copy paper or Kodak (tm) 90% white card in front of the camera and shoot some frames. Use DANCINEC.EXE (tm) to process the frames into a TIF file using the Engineering de-Bayer method that does color interpolation without KCC color matrix. When you use Text mode in the execute menu there is a digital density probe in the graphics preview that comes up if you use the full screen (VESA VBE or DIRECT-X (tm)) display or the external DANGUIVU.EXE (tm) viewer pop-up. You get the probe by pressing [P] on the keyboard, or in DANGUIVU.EXE (tm) pressing the left mouse button (see the help screens in DANGUIVU.EXE (tm)). The value you read for the Kodak (tm) 90% white card or pure white copy paper for the Green pixels can be your guide to getting a value for the EI/ISO, use the normalized Units readout. Since many of the sensors used for Digital Cinema Cameras "top out" or "clip" the Green pixels with an exposure of a white subject at about EI/ISO 160, you can find the ratio of the exposure in stops if the sensor operates at minimum gain since that can give a doubling of the sensor ADC count for each f/ or T/ stop change in exposure. You need to adjust for the "black level" that the sensor is operating at since the sensor has black higher than a count of zero if its working as it should. To measure the sensor black level, cap the lens and turn the room lights off and shoot some black calibration frames, then process the same to a TIF as for the exposed frame using the Engineering mode that does color interpolation without KCC WBC at all (RAW processing for dark green image). Use the probe in the Test graphics preview to read the black level from the TIF made for the Test preview, you should get a reading of about 0.04 to 0.065 maybe for the Green pixels using the normalized Units type. To find the "signal level" you can use something like this formula, signal_level = image_90%_white_level * (1/(1-black_level)) So for example you read 0.5 for the white subject in the exposed frame and you read 0.05 for the black level, you can then make the adjustment for the black level like this, signal_level = 0.5 * (1/(1-0.05)) signal_level = 0.5 * (1/0.95) signal_level = 0.5 * 1.0526 signal_level = 0.5263 If 90% White card is 100% or 1.00 signal for the Green pixels, and that corresponds to EI/ISO 160, and the sensor gain is one T/ stop per doubling of signal, then you can approximate the EI/ISO of the test exposure this way, EI/ISO = 160 * (1/0.5263) = 304 A very rough table can be made this way, Adjusted Signal = 1.00, EI/ISO = 160 (about 12 data bits under white) Adjusted Signal = 0.75, EI/ISO = 213 (about 11.5 data bits under white) Adjusted Signal = 0.50, EI/ISO = 320 (about 11 data bits under white) Adjusted Signal = 0.33, EI/ISO = 484 (about 10.5 data bits under white) Adjusted Signal = 0.25, EI/ISO = 640 (about 10 data bits under white) Adjusted Signal = 0.20, EI/ISO = 800 (about 9.5 data bits under white) Adjusted Signal = 0.125, EI/ISO = 1280 (about 9 data bits under white) Adjusted Signal = 0.1, EI/ISO = 1600 (about 8.5 data bits under white) Adjusted Signal = 0.0625, EI/ISO = 2560 (about 8 data bits under white) Adjusted Signal = 0.05, EI/ISO = 3200 (about 7.5 data bits under white) Adjusted Signal = 0.03125, EI/ISO = 5120 (about 7 data bits under white) Adjusted Signal = 0.025, EI/ISO = 6400 (about 6.5 data bits under white) I included the approximate data bits for each EI/ISO level when a 12bit ADC is used to record 12 bit RAW data as linear in ratio to the T/ stops. If a 14bit ADC is used with 14bit recording then you would add two data bits to each value shown. Because the gamma 0.45 gamma correction used in converting RAW sensor data into monitor viewable images expands the lower tones, you lose about two bits before histogram gaps show up, so you need to SUBTRACT two bits from the 12 bit data values shown to avoid histogram gaps in the graded images. The higher EI/ISO values get expanded more and so can show more histogram gaps when in the graded images you have less than 9 bits of graded data, so for 12bit data histogram gaps may start showing up at EI/ISO higher than 640 maybe, there is some gain in data bits from size reduction, over-sampling of the sensor pixel area vs. the graded image pixel area, and the interpolation of the Bayer pattern, so its hard to give an exact value for when histogram gaps will become a noticeable issue with banding and such but with only 6.5 bits to start with at EI/ISO 6400 it more likely that size reduction and interpolation may not cover all the banding etc. If your sensor clips white for the green pixels higher or lower than EI/ISO 160 you can scale that rough table in ratio. EI stands for Exposure Index and is the arbitrary value you set the light meter to so that you can adjust the lens iris and ND filters used or the shutter angle on the camera to control the exposure. ISO is a standard, like ASA and DIN were for film speeds. Shooting with a True RAW camera, there are no standard ways to process the data, so 18% mid-tone and shadow and highlight detail can be processed with various S-curves giving varied results, you will need to pick a EI number and set the ISO on your light meter with that, saying EI/ISO says that the EI number is in ISO units and is arbitrary rather than meeting the ISO standard documents method of calibration with your light meter. By the way your light meter may not be calibrated to the standard Kodak (tm) 18% Gray card, so you will need to shoot tests and adjust the EI/ISO number you set your light meter to so that you get the right adjusted signal level from your processed True RAW sensor data. EI may also be written E.I. sometimes. Kodak (tm) sometimes rated their industrial film types in EI units rather than ASA units to show that the speed numbers were (inflated) higher than the ASA standard would show for that film stock, meaning that there was no under exposure latitude or safety and that you needed to make careful light readings based on the EI value set on your light meter's ASA selector. Arithmetic ISO units are similar to ASA units, but your light meter may not be calibrated to the ISO standard if it says ASA on its speed selector. Log ISO units are like the old DIN system, which I do not use in this documentation. Q74) I setup my frame processing disks on my main computer, then plug them into my "render farm" computers, but the drive letters do not come out the same, how do I deal with that issue? A74) There is a string replacement command in the utilities menu of DANCINEC.EXE et al. (#5 in its main menu) that can automatically swap parts of the file path "on-the-fly" while the frames are loaded and saved, as well as for the various configuration files used by the programs. Say your "work" disk is N:\ on your main computer you use for grading, and when you move that work disk to the "render farm" computer the OS makes it E:\, you would then go into the utilities sub-menu and select the string replacement command, then enter N:\ as the string to "seek" and E:\ as the string to "replace". Don't just enter N for seek and E for replace since that would swap any time that letter shows up in a file path or file name, you need to include the colon and back- slash in order for just the drive letter to get swapped. The program should then find the right files even though the drive letters may show wrong on the program readouts or prompts (that can show the strings before the swaps are done). If you get any file not found errors, check what string swaps you have setup, and the actual drive letters where the files are and the OS has assigned at the time of running the program, as the OS may change the drive letters depending on what other drives are plugged into the computer. Q75) Someone on CML says that midtone for 18% Gray card should be code 104/255 for 8bpc Rec.709 files, so why do you say to use 118/255 for midtone when using the probe in your programs? A75) For the most part my programs are intended for use in Digital Cinematography rather than "video" use. It seems that video files have overrange headroom going to about 109% or so, and therefore to not have maximum white at maximum code value, as you would in normal BMP files for use on computer monitors. If your DIY film recorder has 8bpc code range of 0 to 255, then you cannot grade you maximum white level above code 255, nor would you want to grade the maximum white level under 255 since you want to use all the ADC bits when you can to reduce banding in the transfer to the 35mm film (DANCINEL.EXE (tm) also uses more than one exposure to "store" more than 255 levels onto the 35mm film, in part overcoming the limits of the 8bpc graphics card and LCD monitor). Search the internet for two documents, cineon1.pdf and cineon2.pdf, as they have information on this subject. In general people call files made for monitor viewing Rec.709, when in fact those files may not be within the specification itself, but rather files that have midtone for 18% Gray subject close to what looks like an equal mixing of the monitor maximum white and minimum black. Standard monitor gamma is 2.4, and standard camera gamma is 0.45, but the camera's have a ISO curve applied and are underexposed to get a more film like headroom and soft clip, so there are no real cameras using the ideal gamma 0.45 curve in normal use. The ideal camera would put 0.18 sensor signal at 0.462 file signal, 0.18^0.45 = 0.462, so 0.462 can be used as the reference code in monitor viewable graphics files as to where an 18% Gray subject should reproduce, but as mentioned the Rec.709 specification included adjustments for out of range signals. Likewise someone on DVinfo.com (sm) said that Rec.709 for 10bpc DPX should have black at 64, midtone at 416, and white at 960. Q76) My computer keep freezing when I try to use your DANCAD87.EXE (tm) program to do grading and make the needed KCC files for processing my frames in DANCINEC.EXE (tm), how do I prevent this problem? A76) It may not be possible to use any USB device while grading with DANCAD87.EXE (tm) or DANCAD3D.EXE (tm) v3.7N because if you press keys on the keyboard by accident while the graphics screen is redrawing Windows XP Home (tm) can develop a system freeze, requiring a power down and re-boot. It is known that this problem happens when USB to SATA harddrive adapters are used, as well as USB to memory card readers are plugged into the USB port even if there is no memory card in the reader. This issue may extend to any USB device such as a USB mouse or USB keyboard. You may need to use a computer that has a PS/2 mouse and keyboard for grading, and not to have any USB device plugged in while doing the grading to avoid this system freeze problem with Windows (tm). This problem may stem from the USB driver using DOS memory at the same time as the graphics card using DOS memory for its BIOS to draw to the screen, and then the keyboard BIOS also trying to use the DOS memory also, and so the computer dies and freezes when you press a key while the graphics screen is redrawing. This issue does not seem to impact DANCINEC.EXE (tm) using the DANGUIVU.EXE (tm) graphics preview, as that uses the Windows (tm) graphics procedures rather than using the graphics cards BIOS. Its mostly an issue for making the KCC files in the Grading control screens, so using another computer to do the grading may be one way to avoid this issue. At some point later if I am able to update the grading screens to run under Windows (tm) as a GUI program, then this issue should not be a problem, but I do not know if or when I can do that code port, so for now (Aug 2011) using a PS/2 keyboard and mouse may be needed. Q77) I notice that the readout in DANGUIVU.EXE (tm) at the bottom of the screen is cut off by the task bar in Windows 7 (tm) and or the window draws then redraws on first opening, how do I fix this? A77) When you select the "text size" in Windows 7 (tm) or maybe also other Windows (tm) versions, the number of pixels high that the blue bar at the top of the window is high (width) changes, as far as I can make out the sizes are: Small text = 27, Medium text = 28, and Large text = 33. Under Windows XP Home (tm) it seemed that the small text size was 22 pixels. The way you can tell if the number you enter in the graphics setup for DANGUIVU.EXE (tm) is right or wrong is to push the window off the side of the screen than draw it back using the mouse on the top blue bar, the window active area will go black and show some numbers, the window area numbers should equal the preview image size numbers when you have the top width set right, and you should be able to see all of the readout text at the bottom of the preview image without any black area between the bottom of the preview image or gray background and the bottom part of the window frame border. Q78) When I use the full screen Direct-X (tm) graphics display under Windows 7 (tm) only a strip at the top of the screen is visable, how do I fix this? A78) I do not know how to fix this Windows (tm) issue, it works under XP Home (tm), so you will need to use DANGUIVU.EXE (tm) or some other external program for now if the Direct-X (tm) full screen does not work on your computer. The Direct-X support is part of the graphics library that came with the compiler I am using, not something I wrote, so what the issue is with it not working on some computers, but working on others is yet to be figured out. Q79) I'm using the presets and want the image to come out full size rather than be resized to the size I am getting how do I fix that? A79) Press [R] key when you are in the execute menu which is #3 from DANCINEC.EXE (tm)'s main menu. If you are using a RAW camera without header data, you need to enter the image size in the camera models menu #1 then #3 from the program's main menu. Q80) I notice under Windows 7 (tm) that the image tones do not match between what I see in DANGUIVU.EXE (tm) and the windows graphics file viewer when I open the result file, how do I fix that? A80) I don't know, its Windows (tm) does anything work right? I am using the windows pixel write in DANGUIVU.EXE (tm) you would think that it uses the same values in the video board when the Windows (tm) viewer is used, why would you scale a BMP file, its a 1:1 data transfer? For DPX/CIN files you will get things not looking right because viewer needs the display LUT to match, which does not involve Windows (tm) but rather the display programs. Q81) The program crashed after I set the graphics size for the DOS VESA graphics to the same size as my source RAW file, is that a bug? A81) You are in the wrong menu, the RAW image size gets set in the camera model menu, #1 then #3 from the main menu, the video mode size gets set in #4 then #2 from the main menu. The video mode size for full screen graphics must be an exact match to the video modes supported by your graphics card and the monitor in use, otherwise there is an error produced. If the source RAW image is processed larger than the full screen size, the preview image will be resized to fit the screen size. --- KNOWN ISSUES AND PROBLEMS Known issues and Problems mostly have to do with things that are out of my hands because they are due in or because part of the OS or compiler does not always work as one would hope, rather than obvious "bugs" that I might be able to fix. For the most part I do not ignore programming "bugs" that I am able to see on my computer, and so you should report anything that is not working. Do not assume that because this is developmental code released for "Beta Testing", I know that something is not working, for the most part if I see a programming error I fix it soon after I find it, or at least when I am able to. Sometimes things that seemed to be working get "broken" because of changes made to add new features, so there may be things that need attention that I don't know about. If you want to make use of the programs it is in your own interest to tell me if there is a problem and to continue to answer my questions about any details of the issue or how some attempt at resolution turns out. 1) If you make a set of BMP frames with DANCINEC.EXE (tm) and then open that set in the third party freeware program VirtualDub (tm) to watch the frames play at sync speed, and then try to re-generate the frame set with DANCINEC.EXE (tm) open in another "window" you will get a file I/O Error message. The reason for that is that Windows (tm) marks the BMP frames as in-use by VirtualDub (tm) and so makes an I/O Error when DANCINEC.EXE (tm) tries to overwrite the BMP frames with new versions. The way to prevent that problem is to CLOSE VirtualDub (tm) BEFORE you execute the output of new BMP frames with DANCINEC.EXE (tm). This is an easy mistake to make, and can be confusing. The same problem can happen if any other program has files in use while you are trying to make updated ones with DANCINEC.EXE (tm), so be sure to close other programs, or at least their having files open while you try to overwrite the same filename with a new version. My programs will report I/O Errors and other Error messages if there are any file sharing conflicts as they have no way to "share" files, only read or write them. 2) If you use DANCAD87.EXE (tm) to make a sound track mix for a shot or shots you are editing, and you open the mix-down *.WAV file in Magix (tm) then make some changes with the editing or mix down settings in DANCAD87.EXE (tm) and then try to reload the new *.WAV into Magix (tm) you can get problems in Magix (tm) because Magix (tm) may not detect that the length of the *.WAV file has changed etc. Magix (tm) makes some files that contain the graphics of the waveform etc. in the same directory as the *.WAV file its is loading, if you manually delete ALL the files in the Mix down folder in the Project folder (of the folder Structure used by my programs) before you do the new mix-down, and close Magix (tm) and then restart it, that should clear Magix (tm) from "remembering" the length and other facts about the *.WAV file, even though the new mix-down has the same filename. You might also be able to re-name the mix- down file and leave Magix (tm) open, but load the new mix-down under a new project name in Magix (tm). This may also be an issue in other WAV editors, so if you get errors in the playback of your mix-down it is something to look into before you think there is a problem with my programs, that is just delete all files with the same name as my mix-down file then re-generate the new mix-down before you load it again. The mix-down is in the /M folder of the /Rxxx folder in the project folder, for example C:/P0001/R000/M/*.WAV, most of the time you will just be using reel zero for the full length playback mix-down, if you make reel by reel mix-downs then they could be in folders R001, R002, R003, R004 and so on. 3) Windows (tm) XP Home (tm) can have problems with setting the higher resolution video modes when using DANCINED.EXE (tm) the 32bit DOS version full screen display. Unlike ME (tm), XP Home (tm) seems to prevent direct access to the VBE 3.0 VESA BIOS and sets the video refresh to rates unsupported by the monitor and also prevents my program from being able to set the refresh to the right rate, at least as far as I can tell so far. This may be do to having at some time turned on the composite video output on your video board, and maybe that is "stuck" on because you updated your video board or driver. There may be no direct way to fix that issue other than a complete new install of everything, even that may not help. One thing that should work is to use a computer with the Windows ME (tm) full GUI install, since Windows ME (tm) does not seem to interfere with the access to the video board like Windows XP Home (tm) seems to. But to use Windows ME (tm) you need to have ALL the harddrives in the computer formatted FAT32. There are hardware PATA ribbon cable (from motherboard) to SATA drives adapters that don't need a software driver and those may let you use SATA drives without 640K DOS memory dual use conflicts (this is being looked into). Anyway if you format your HDD FAT32 you can use them under almost any OS, Windows ME (tm), Windows XP Home (tm), Linux (tm), and others so it makes your work and archive disks have more universal access and there is less need to copy files from one disk format to another to use the files on various computers such as going from PC to MAC etc. Search for the freeware utility fat32format.exe to overcome the crippling of the format command in Windows XP Home (tm). If you have problems with the VESA full screen display, you can select the DANGUIVU.EXE (tm) display option and get up to full desktop width and 13 pixels less than full desktop height. 4) The Windows Console (tm) version DANCINEC.EXE (tm) uses an older 32bit compiler with Direct-X (tm) library for its graphics modes. I have to date not been able to get the source code for that graphics library to compile so I can see about making updates to its code. Because of that you may not be able to get the graphics test display to show on 16:9 monitors, or at least not with the right aspect ratio on the pixels. If that is the case the best solution would be to plug in a 4:3 monitor for now. Also the Direct-X (tm) version seems only to work up to 1600x1200, where as the 32bit DOS version works up to 2048x1526 on video boards that support that in VESA modes. So I don't know if you can get 1920x1440 or 2048x1536 graphics display under the Direct-X (tm) library for graphics. If future versions of my programs can overcome these limits has not be determined at this time. If you are using Windows ME (tm) and your video board supports 16:9 video modes in its VESA VBE 3.0 BIOS, then the 32bit DOS version DANCINED.EXE (tm) may be able to display those 16:9 video modes if your monitor supports them. CRT monitors that support up to 2048x1536 should be able to display various aspect ratios, and have width and height adjustments for the image display. If you have problems with the VESA or DIRECT-X (tm) full screen display, you can select the DANGUIVU.EXE (tm) display option and get up to full desktop width and 13 pixels less than full desktop height. 5) DANCAD87.EXE (tm) seems to have some kind of conflict with a USB HDD adapter I have on my system under XP Home (tm), when both are in use at the same time the system freezes and cannot be revived even with [Ctrl]-[Alt]-[Del]. I read somewhere that SATA BIOS can use DOS memory, if so that may be an issue, but it seems to happen more when DANCAD87.EXE (tm) is in graphics mode and graphics also use DOS memory so there might be something to that, not that I can make up for the OS, Video BIOS, and SATA BIOS not sharing the DOS memory in a way that avoids crashing the computer. If nothing else helps, copy the files you need to work on onto an EIDE drive and use that, with the USB and or SATA drives unplugged. Such system freeze may also happen if you are using the mouse or keyboard during a graphics refresh of the screen, maybe because of too many interrupts, again such OS or motherboard and CPU faults that might contribute to such issue don't have much of a solution other than to use another computer, change the hardware around, or use another OS version. See point 14 below. 6) If you have DANCINEC.EXE (tm) open and running, and you open and run DANCAD87.EXE (tm) in another window or full screen then try to use the Grading command in DANCAD87.EXE (tm) the video board VESA BIOS probe to find the video modes supported goes very slow and takes several seconds. To avoid that issue, close all copies of DANCINEC.EXE (tm) before you run DANCAD87.EXE (tm). Also I would not have both programs making graphics full screen displays at the same time, your system will probably freeze, even though Windows (tm) should let you have two programs write to full screen with one running in background and the other in foreground, what happens is that Windows (tm) pops you out of the first program to write to the screen for the second program, then cannot let you back into the first program with the screen restored right, you may get a black screen, from which you may be able to [Esc] out (or use other keys such as [Return]) but you would be working blind, and if you are working blind you may press the wrong command key and get into a prompt that asks you to enter something but you don't know what and it just keeps beeping at you with a black screen on the screen. To avoid going nuts with such Windows (tm) issues, its best to just run one program at a time, and not to pop out of my programs while you are in parts that use the graphics full screen display, if you need to pop out to the desktop or another program, only do so when my programs are in the 80x25 char text screens (blue with white text), which Windows (tm) seems to have fewer problems restoring after a pop out. 7) Don't try to run more than one copy of any of my programs in the same folder as they use temp files and those files would have the same filenames, and would try to overwrite each other. If you need to run two or more copies of my programs at the same time, you should run them each from their own folder. 8) If DANCAD87.EXE (tm) crashes or your system does a freeze or you halt it manually with [Ctrl]-[Alt]-[Del] you may need to manually delete a file called DANCADXX.RUN in DANCAD87.EXE (tm)'s folder before it will be able to be run again. DANCADXX.RUN is made to try to prevent users from starting the program twice in the same folder, but if the program does not exit in the normal way, it does not get deleted. Running DANCAD87.EXE (tm) under ME (tm) may show fewer issues than later OS. 9) The Windows GUI (tm) version DANCINEW.EXE (tm) does not have a full screen graphics preview, but rather uses the DANGUIVU.EXE (tm) program by default for its "digital density" probe feature and test preview etc. The GUI version also seems to have problems after a "file not found" error happens, why that should be and not in the Console version probably has something to do with Windows (tm), maybe, or who knows what, since they share the same source code. I would avoid using the GUI version if you can, and if you can't be very careful that you do not enter the wrong filenames and such. Also some of the keyboard UI related features that work in the DOS and Console versions may not work in the GUI version, those are not "bugs" in my code as such, just the compiler and Windows (tm) may not support some keyboard functions the same under GUI. If the GUI version starts to act odd, quit the program and re-boot the computer and see if that helps, if not, then re-install the program in a new folder from the DANCINEC.ZIP (tm) file. Be sure to quit DANGUIVU.EXE (tm) or the external graphics viewer program you select after looking at the test frame, since you do not want to close DANCINEW.EXE (tm) and leave the graphics viewer program open, which could cause problems the next time it needs to open. 10) In the Windows Console (tm) version the [Ins] key may not make the cursor change from small to large to show that your are in insert or overwrite data entry modes, like it should in the DOS version. That seems to be due to some compiler issues with how the console works under Windows (tm) and I'm not sure if there is a solution to that, for now you just need to keep in mind the [Ins] state when you are entering numbers and filenames etc. 11) Although DANCINEC.EXE (tm) may let you enter long filenames, the program is not designed to be used with long filenames at the time I am writing this, and in the DOS and Console versions the file directory only shows 8 char folder or filenames plus 3 char extensions. To avoid problems only use 8 char folder names, 8 char filenames, and 3 char extensions. Do not use double periods in file names, there should just be one period between the filename and its extension. 12) Some cameras may call RAW files DAT files, if you get source file not found, you may need to override the source extension with the prompt in the "generic" camera type option #1 in the camera model setup menu. Otherwise, rename your RAW data source files RAW all the time. 13) Do not open the result file with some other program, then try to make a new test frame or process a new result frame set. If you do that you will get serious I/O Errors because my programs cannot write to a file that is open in some other program, they just detect that as an I/O Error. This is a very easy mistake to do when you have several programs open at once, the same problem comes up when you have the WAV mix file made by DANCAD87.EXE (tm) open so you can listen to it, and you make changes and do a re-mix pass, the result file cannot be re-saved because you have it open in another program. So if you open any of the result files in a viewer etc. be sure to quit that program be for you re-generate the result file so the new result file can overwrite the older one. 14) It may not be possible to use any USB device while grading with DANCAD87.EXE (tm) or DANCAD3D.EXE (tm) v3.7N because if you press keys on the keyboard by accident while the graphics screen is redrawing Windows XP Home (tm) can develop a system freeze, requiring a power down and re-boot. It is known that this problem happens when USB to SATA harddrive adapters are used, as well as USB to memory card readers are plugged into the USB port even if there is no memory card in the reader. This issue may extend to any USB device such as a USB mouse or USB keyboard. You may need to use a computer that has a PS/2 mouse and keyboard for grading, and not to have any USB device plugged in while doing the grading to avoid this system freeze problem with Windows (tm). This problem may stem from the USB driver using DOS memory at the same time as the graphics card using DOS memory for its BIOS to draw to the screen, and then the keyboard BIOS also trying to use the DOS memory also, and so the computer dies and freezes when you press a key while the graphics screen is redrawing. This issue does not seem to impact DANCINEC.EXE (tm) using the DANGUIVU.EXE (tm) graphics preview, as that uses the Windows (tm) graphics procedures rather than using the graphics cards BIOS. Its mostly an issue for making the KCC files in the Grading control screens, so using another computer to do the grading may be one way to avoid this issue. At some point later if I am able to update the grading screens to run under Windows (tm) as a GUI program, then this issue should not be a problem, but I do not know if or when I can do that code port, so for now (Aug 2011) using a PS/2 keyboard and mouse may be needed. --- RUN TIME ERROR MESSAGES Here is a list of run time error messages that you might see some of if the program crashes. Most of the time it would be divide by zero, 200, or memory overflow issues, 216, if you see any at all. If you do not see on of these messages on the screen when the program crashes, then it may be a Windows (tm), Video board driver or BIOS, or mouse driver issue rather than a logic bug in my programs. Anyway, if you see any of these numbers on the screen you can look up what they mean here, these are made by the compiler run time library and not my code. These are only for my current 32bit programs, DANCINED.EXE (tm), DANCINEC.EXE (tm), DANCINEW.EXE (tm), and DANCINEL.EXE (tm), the other programs use other development tools so have their own unique run time error codes. It may be a good idea to re-boot your computer after a program crash, and to see if you can avoid the issue when you run the program again, be sure to check that filenames and values used in setup are valid. If you get recurring error codes from this list you should try to tell me what you are doing that results in their being displayed so I can know about the issue. For the most part in normal use you should never see any such error messages and the program should never lock up your system because of errors in my code, problems with Windows (tm) et al. crashing your computer are for the most part out of my reach. If you get a system freeze, try pressing [Ctrl]+[Alt]+[Del] at the same time and see if "task manager" pops up, if so you may be able to close my programs using "task manager". Some program operations take quite a long time with large image sizes, so you can check to see if the program looks Busy with "task manager", you may see the memory use getting larger and smaller telling you something is going on, if the memory use keeps getting larger until the system crashes then that may mean a memory "leak" or there is not enough memory free. For the most part you should not run any programs in background while my programs are running and you should not run more than one of my programs at the same time as they cannot share files and hardware, in particular don't try to pop out of my (DOS and Console) programs while they are in graphics display as Windows (tm) may not let you back in (or may show a black screen when it does), when you are in the text screens with white text on the blue background you may be able to pop out with the Windows (tm) key and get back in without issues. Code - Meaning 1 - Invalid function number 2 - File not found 3 - Path not found 4 - Too many open files 5 - File access denied 6 - Invalid file handle 12 - Invalid file access code 15 - Invalid drive number 16 - Cannot remove current directory 17 - Cannot rename across drives 18 - No more files 100 - Disk read error 101 - Disk write error 102 - File not assigned 103 - File not open 104 - File not open for input 105 - File not open for output 106 - Invalid numeric format 150 - Disk is write protected 151 - Bad drive request structure length 152 - Drive not ready 154 - CRC error in data 156 - Disk seek error 157 - Unknown media type 158 - Sector not found 159 - Printer out of paper 160 - Device write fault 161 - Device read fault 162 - Hardware failure 200 - Division by zero 201 - Range check error 202 - Stack overflow error 203 - Heap overflow error 204 - Invalid pointer operation 205 - Floating point overflow 206 - Floating point underflow 207 - Invalid floating point operation 208 - Overlay manager not installed 209 - Overlay file read error 210 - Object not initialized 211 - Call to abstract method 212 - Stream registration error 213 - Collection index out of range 214 - Collection overflow error 215 - Arithmetic overflow error 216 - General protection fault 217 - Invalid operation code 227 - Assertion failed 300 - File IO error 301 - Non matched array bounds 302 - Non local procedure pointer 303 - Procedure pointer out of scope 304 - Function not implemented 305 - Breakpoint error 306 - Break by Ctrl/C 307 - Break by Ctrl/Break 308 - Break by other process 309 - No floating point co-processor 310 - Invalid Variant type operation --- ADDITIONAL INFORMATION Please visit my Web site On-Line at: http://www.DANCAD3D.com/ for additional information. Please report bugs, mistakes, or other problems with this document or the programs, see SECTION: 8 at the current On-Line version of my Web site for current instructions. Since this program is in very preliminary development stages, you should report any issues you find, as for the most part I do not knowingly distribute programs that have obvious serious issues such as not running at all (under the recommended OS) and so forth.

This copy of this page was compiled on or around: Y2011.M11.D23, you might check the "On-Line" version, or come back later, to see if there is a newer compile.