@ldigirol, sorry for the belated response, but I’ve been messing with just this, using ICC profiles. There was a discussion in another thread recently that cast concern on traditional WB compensation in the context of being just another arbitrary channel modification, where the chromatic relationships are disturbed. E.g., if you aggressively apply a curve, the colors will also change visibly. A white balance multiplier is in that category, and the camera numbers can be especially aggressive.
So I conducted a little experiment with a previously generated camera profile produced on a white-balanced target shot. I re-generated the profile, but this time i used another raw development of the target shot, one that was only demosaiced and not white-balanced. You can regard the effect of the two profiles here, as well as read some of the surrounding discourse:
The color difference is marked.
I’ve re-developed some of my sunlight-based images with this new profile, and the starting saturation is such that it compels me to not have to apply saturation later in the processing. Effectively, the white balance correction takes place in the color conversion from the camera profile to the working profile, as what @gwgill called a ‘chromatic transform’.
Now, this is probably a challenging thing to do with ‘regular’ software, as most desperately want to white balance before demosaic. I use my hack software, which lets me start with the ‘really raw’ bayer mosaic, so I can add operations in any order that suits my fancy. I also don’t have the capability to use DCP profiles, so I can’t speak to whether or how this concept would apply in that workflow.
But yes, my investigation to date would lead me to answer Yes, it would be better to use camera profiles to compensate all camera chromatics, including white balance.