Martin Weberg wrote:
I could try this. Stripping the FOGRA39L data of all combinations using K except for K=0. Then create a CMYK profile. Or should I also strip the entire CMYK_K column creating a CMY-profile?
Yes, the latter. The former won't work, since various things will fail if the device response doesn't occupy space in CMYK and PCS.
Then I guess there is a possibility with an iterative process finding the closest match (using i.e. dE2000). Like, feeding icclu with a seed (educated guess) sub-set of the device color and then compare the resulting Lab color with the reference Lab color. The process proceeds exploring nearby sub-set device colors and chooses the "path" resulting in minimum dE. There is a possibility with different results depending on the seed.
Yes, if speed is no object, then it's possible to write inversion code with different objectives.
Studying marine biology I did a simulation lab using this sort of thing. Just can't remember what the process/algorithm is called. Maybe, you know a way of doing this?
It's either root finding or minimisation of an objective function. Naturally there are both these sorts of routines in Argyll itself :-) Graeme Gill.