Hallöchen! I followed http://blog.pcode.nl/2010/06/28/darktable-camera-color-profiling/ to get an input colour profile for the NEX-7. I have already sent the RAW and the reference file to the DT developers, so I think they will convert it to an enhanced colour matrix when they have the time. However, this profiling was a daunting experience the observations of which I'd like to share:
1. Calling colprof without the "-u" parameter means that the white patch (with L=90.45 in my case) is converted to L=1 after the colorin module. This makes the image brighter. However, the standard colour matrix doesn't do this. Additionally, the base curve cannot be applied anymore because the image gets way too bright then. 2. Exporting with "colorimetric (absolute)" means that the white patch becomes yellowish. All other colour intents work properly as far as I can see. I have "always littlecms2" activated. If I deactivate it, there is never a colour shift. 3. The quality of the ICC seems to be suprisingly poor. If I apply the ICC to the original calibration image, I expect the values on the patches to correspond with the values in the reference file. For the patch "L19" (deep blue), the reference file says: (in Lab) 12.62 39.19 -65.00 In Darktable, I get (after exposuring to the white patch L=90.45 and chcking white balance): 18.6 19.3 -51.3 (matrix) 17.5 18.4 -49.1 (shape+matrix) 20.9 17.1 -51.2 (Lab cLUT) 23.7 11.1 -47.2 (XYZ cLUT) If I use dcraw for applying the profile with "dcraw -4 -T -o profile.icc DSC00001.ARW" (measured in DT): 15.3 28.9 -56.0 (for ALL profiles) Yes, I really double-checked that, the given profile file didn't matter. WTF? If I give "-o 0" I get much worse values, and without -o at all (i.e. standard Adobe colour matrix) I get 15.2 28.6 -55.6 (Adobe colour matrix) I find all these figures disappointing. And the LUTs performed worst. And it is noteworthy that dcraw performes different from DT with the same profile files, and better. So is it possible to get a reasonable (i.e. 5-10 times closer) enhanced colour matrix somehow? If so, how? Tschö, Torsten. -- Torsten Bronger Jabber ID: [email protected] or http://bronger-jmp.appspot.com ------------------------------------------------------------------------------ This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev _______________________________________________ Darktable-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/darktable-users
