Lars Borg wrote:
This can easily lead to cases where a pair of profiles have no intents in common: Profile 1 TRC relative only -> profile 2 perceptual only. This is a common conversion from display to print. There is no common intent, so conversion should then not be allowed.
That makes the assumption that the user only indicated one intent to cover the two profiles involved. Yes, many CMM's dumb it down to one intent, creating this issue, while making it simpler for the user and keeping them in dark as to what's going on. (This is all part of the deeper problem of casting the illusion that there's real gamut mapping going on here, something that's difficult to impossible to achieve using pre-calculated source and destination profiles where the profiles are meant to implement the gamut mapping.) The lack of separate intents created the user "surprise" problem in the first place with the use of absolute intent in Photoshop. What users most often wanted was absolute intent for the input printer profile, and relative colorimetric for the display profile. Yes, having fall-backs makes it all easier to use in practice.
Also to consider: BPC is not colorimetric, so relative + BPC should be classified as perceptual.
In my view BPC is a lightness axis only, source & destination tailored gamut mapping. So BPC + relative = perceptual mapped light axis + relative colorimetric in every other way.
And is saturation + BPC == perceptual? If so, all intents become perceptual if BPC is on, so all combos are valid!
Since BPC only scales the lightness axis, BPC + saturation = saturation more accurately tailored to the particular source space. Graeme Gill.