Alastair M. Robinson wrote:
Anyway, my colorimeter can capture samples under D50 or D65 (possibly others too, haven't checked yet). Prints produce using Perceptual transform with the D65-generated profiles are noticeably more pleasing that the D50-generated ones.
I don't really understand why.
D65 may better represent your actual viewing conditions, or it might simply push the color in a direction that tends to compensate for other inaccuracies.
If I scan an IT8.7/2 target, profile the scanner, then do an absolute colorimetric transform to the printer profile, I get a pretty good match (within the limitations of the scanner) with the D50 profile and a very yellowy print with the D65 profile.
Right, because you've applied the D65 viewing conditions to one profile, rather than both. If you are going to create a profile with a non standard measurement illuminant (say D65), then you need to apply similar conditions to your other profile, if you are going to link them together with absolute. If you don't, you will get a white point shift that you probably don't want.
Since you can't really control the scanner illuminant, it's hard to see a scenario in which absolute colorimetric linking makes sense.
The reason for my orginal question was that I was wondering if patching the D65 profile with a D50 white/black point would result in a profile usable for both perceptual/rel.col and abs.col - if I can avoid having two versions of every profile lying around it would be nice :)
Measuring under something closer to your actual viewing illuminant is a very reasonable thing to do, as is using such a profile with relative colorimetric (when the source white point is different), or using absolute colorimetric (when the source white point should be emulated). It's harder to justify a scenario in which you want to measure under D65, but use it in a pseudo-absolute mode in which the "absolute" white point is a bradford transformed D65->D50.
My experience with absolute mode is that it is only useful when doing proofing, where both the source and destination are going to be viewed under the same illuminant.
I guess the question is, what is the purpose of using absolute colorimetric mode between a scanner and printer ?
If there is a compelling answer to the question, then perhaps what you are missing is a D65 measured reference file for the test chart used to characterize the scanner ?
Graeme Gill.
