Milan Knížek wrote:
The perceptual (and saturation, too) intent requires that both source and target gamuts are known when profiles are created and this intent is supported only by LUT profiles (not the shaper/matrix ones).
I think that's a reasonable statement (and true when applied to Argyll's profile), although I'd clarify by saying that only the destination profile has to be a LUT type, since that is where the gamut mapping is done. Others might disagree though (particularly those who are in the "ICCV4 PRMG works" camp), since by distributing the gamut mapping between source and destination with an agreed PCS gamut in between, (what I would call) saturation intent is supported in a mix and match fashion for LUT profiles. Note that this approach can be employed with V2 profiles, but you would only expect it when source and destination profiles created by the same software.
Which implicates that if one wants to use the perceptual intent for displaying images on the monitor, one must create several monitor profiles for images in working spaces bigger then the monitor gamut (AdobeRGB, ProPhotoRGB, etc.) and possibly also for the printer profile for soft-proofing.
These monitor profiles would behave the same way for the colorimetric intents, but will provide different results for the gamut mapping intents (perceptual, saturation).
Using Argyll's approach, yes. Graeme Gill.