I saw a hack here on using Canon color science on Sony RAW files: https://youtu.be/4bnAimJhwKs?si=PZ_0zzcVIo-T7a6r
I want to try the same in Darktable.
I’ve found DCP profiles for different Canon cameras, but no ICC profiles. I haven’t been able to convert the DCP profiles to ICC profiles with any of the tools I’ve found for the purpose online.
Case is I have a Sony Nex 3n, and I don’t really like its color science, so I want to experiment with alternatives.
Using the wrong matrix is hardly colour science. You can use an instance of color calibration as a channel mixer and achieve any matrix twist you want, or you could use rgb primaries, which is also backed by matrix multiplication.
If you want the Canon curves, try the various curve tools in darktable.
Will I be able to achieve the color separation in skin tones from yellow in this way? I am very much a novice to shooting and editing in RAW, but I hope to find a way to deal with the Sony skin tones, which have so far dampened my enthusiasm for the older Sonys…
You are sure there are no other issues with your edits that could explain the problem with skin tones (e.g. white balance)?
Camera profiles are dialed to the specific spectral response of the camera model, so using them interchangeably across brands does not work well. What I’ve found is that recent cameras of a same brand may have similar spectral response; I’ve used Nikon D7000 profiles on Z 6 images and things came out okay. You can easily see such similarities and differences here:
https://glenn.pulpitrock.net/openfilmtools_ssf_plots/
Look at the two Canon plots, and the two Nikon plots; note the similarities in the brands, and the differences between brands. Really though, using another camera’s profile on your camera is a janky sort of business…
If you want to deal with skin tones, the best way I’ve found is to make a profile for your camera with spectral data, trained with a color patch dataset oriented to skin tones. I have spectral data for a bunch of cameras, but no dataset for yoru camera. I do have a skin tone dataset, the RIT Lipppman 2000 collection, and I’ve made a profile trained with it for my camera. Difference is notable, first image, Disney princess, Nikon Z 6, regular matrix profile:
Same capture, spectral LUT profile trained with the Lippman data:
She’s a bit warmer in the second image…
