match JPEG colors in Darktable

Aargh. Then the least-tedious solution would be manually changing WB to “as shot” for each image. I wonder if I can use LUA scripts for this (probably not).

I am surprised that auto-detection in color calibration gives me better results for some images with WB “as shot”. But it is image-dependent, occasionally OK, sometimes not.

I think that the best and most general solution will be a matrix, as suggested by @Donatzsky, because WB only has 2 degrees of freedom, while the matrix has 3x3-1=8. After all, this is about the sensor, not white balance.

It is my impression that this issue is highly relevant in general for a lot of cameras. Cf this topic, where unfortunately I was not able to obtain a practical solution:

Again, I am fine with getting a color chart and doing the legwork, but it is unclear what I would do with images afterwards to get the color matrix into Darktable. I am under the impression that they are no longer added.