4
\$\begingroup\$

My understanding is that relative intent just clips the out of gamut colours and that perceptual intent scales (compresses?) the colours to make the out of gamut colours fit. It seems to me there is no processing necessary in case there are no colours out of gamut at all.

Yet, when soft proofing in Lightroom for example, the image visually changes upon selecting relative or perceptual intent, even when there are no colours out of gamut. I would like to understand why. (Probably my understanding of the colour conversion process is too simple.)

Improve this question
\$\endgroup\$
0

3 Answers 3

Reset to default
2
\$\begingroup\$

It's quite common for images that are entirely in gamut to soft proof and print differently between Relative and Perceptual Colorimetric.

Relative Colorimetric Intent is well defined for all in-gamut colors by the ICC. However, Perceptual Intent for printer profiles is not. It's up to the creator of the profiling software. Colors that are in gamut and accurately printed using Relative Colorimetric intent are almost always shifted when using Perceptual to allow room for compressing highly saturated colors. Also, vendors of ICC profiles often increase saturation so that a print appears more colorful. Both of these effects can be seen in the OEM ICC profiles for the Canon Pro1000 printer with glossy paper (Canon PRO-1000/500 Photo Paper Plus Glossy II A).

Canon Pro1000 a* Rel. v Perc.

This charts shows the actual printed a* (from Lab* where L* is fixed at 60 and b* at 0 while a* is varied from 0 to 100, The printer's gamut is limited to approx a*=80. The printed a* in Perceptual is more saturated (44 v a requested 40) below 70 but reverses and becomes more compressed when nearer the printer's gamut limit. For instance when requested to print a*=80, which is right at the printer's gamut boundary and printable in Relative Colorimetric, compression reduces the printed a* color to 73. However it prints at 80 using Relative Colorimetric.

So, as we see here, there will be a discernible color shift for many colors well even when those colors are well inside the printable gamut.

Improve this answer
\$\endgroup\$
1
\$\begingroup\$

The relative intent moves out of gamut color to be w/in gamut. So if you have some blues which are out of gamut and some blues that are just w/in gamut they may all wind up being the same (colors/areas "pack up").

The perceptual intent will also move the out of gamut colors to be w/in gamut, but it also moves the in gamut colors so that they remain perceptually different (i.e. retains a blue gradient).

If the colors are truly in gamut, then you should see no difference between the intents and the histogram should not shift. I suspect you do not have the softproofing out of gamut warning enabled (the upper rt corner of the histogram).

enter image description here

Improve this answer
\$\endgroup\$
2
  • \$\begingroup\$ I do have that option turned on, while having my printer's colour profile selected. That way, all the pixels that are out of the printer's gamut are marked red. Some pictures really don't have any red on them, but there still is a change between perceptual and relative intent. Could it be that there are just a few pixels that are unnoticeable cause they are not clustered together? And LR then processes the whole picture differently based on selected intent and only to push those few pixels in gamut? Is there a way to check for that? Like a flag that says if there are out of gamut pixels? \$\endgroup\$
    – Ralph
    Commented Jul 27, 2020 at 19:54
  • \$\begingroup\$ @ Ralph... does the histogram shift? My guess is that you are not zoomed in far enough to see the out of gamut pixels/warning \$\endgroup\$
    – Steven Kersting
    Commented Jul 28, 2020 at 4:03
0
\$\begingroup\$

Simple answer: Gamut Mapping is different between the two rendering intents. Relative is best if you are not sure the gamut mapping for perceptual is helpful. No scaling going on there at all, so that is optimal when the file is in the gamut of the display

Improve this answer
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.