I'm trying to discover whether a 'trick' I found is a reliable method for dealing with out-of-gamut when soft-proofing.
Down to the 4th picture is just background info of what I'd tried unsuccessfully so far - so skip for TL:DR.

This is a follow-up to Colour Workflow A - Z in my ongoing saga of trying to get things right when I send it to be printed.

Having successfully recalibrated my workflow in Photoshop so I'm not sending out of gamut pictures to the printers & receiving badly-posterised results back, I now check using soft-proofing with the gamut warning on, so I can check I'm not exceeding limits.

How do gamut warnings in LR soft proofing help? works through the how & why of gamut checking [& I also posted examples there of what you get back from the printers if you ignore them.

So, going back through my existing work, getting ready to have more proofs printed so I can test I got it all right so far, I have been working at reducing saturation in some pictures that were clearly far over the top.

These are just small screenshots from Photoshop

So I try to pull back the purple/magenta in Camera Raw until the gamut warning disappears. This, of course, reduces the levels across the whole picture & is deeply unsatisfactory.
Alternatively, I play with the curves.. frankly making a mess of the whole thing...

Then I discovered a trick - or I think I discovered a trick & I really hope I'm not fooling myself & heading for another expensive disaster...

Edit > Convert to Profile... & use the print shop's own ICC profile for the medium I'm going to print on (which is an RGB profile, not CMYK).

It doesn't come out identical to my input, but it does rather pleasingly compress my out-of-gamut areas into known bounds without affecting the rest of the image, & I now have zero warnings & no horrible posterisation.
I'm a sound engineer by trade, so the idea of audio level compression is one I'm very familiar with. It feels like I just 'discovered' colour level compression.

Have I fooled myself - or is that what I'll get back from the print shop?

(I am, of course, working on a copy, I still have the original untouched version to use next time.)

  • \$\begingroup\$ VTC as 'unclear', with no explanation... how am I to clarify? \$\endgroup\$
    – Tetsujin
    Jun 19, 2017 at 16:47
  • 1
    \$\begingroup\$ Sorry, meant to comment, had to walk away from computer at work for a moment. This seems more like a open invitation for discussion. I think my problem with the question is that this requires knowledge of your "saga", so the question feels like it's an episode in a longer discussion, much like a multipart blog post inviting discussion, rather than a concrete question. \$\endgroup\$
    – scottbb
    Jun 19, 2017 at 16:57
  • \$\begingroup\$ @scottbb - I was referring back, for context (I was taught to always show your workings ;) but the question itself is self-contained. Essentially "I think I found something, is it right?" \$\endgroup\$
    – Tetsujin
    Jun 19, 2017 at 17:11
  • \$\begingroup\$ I've reread your question (and the linked ones) a few times, and since I can't clearly spell out specific problems, or what I would suggest to improve it, I have concluded that the problem is me. =/ Retracting. =) \$\endgroup\$
    – scottbb
    Jun 19, 2017 at 21:42
  • \$\begingroup\$ I think it would help if the question title described the thing you are doing, rather than your discovery of it. \$\endgroup\$
    – mattdm
    Jun 20, 2017 at 7:19

2 Answers 2


This is how color management is designed to work. When converting out-of-gamut colors into a new color space, both the Perceptual and Relative rendering intents put those out-of gamut colors at the outer-most edge of the target space's gamut.

Relative intent only affects those out-of-gamut colors, while Perceptual compresses the image's entire color palette until all out-of-gamut colors fit in the target space's gamut.

You will indeed get back from the print shop what you see on screen with this technique (at least as close as color management can get to matching monitor colors to printed colors, provided their printer and your monitor are properly calibrated and profiled, obviously).

  • \$\begingroup\$ That's what I thought - but it does seem strange there's no 'soft-fix' for the soft-proof. Staring at out-of-gamut grey bits, or hard-mapping to a printer profile... when adding a 'compression' layer to any desired output profile could do the job in 2 seconds just seems a bit counter-intuitive. \$\endgroup\$
    – Tetsujin
    Jun 19, 2017 at 16:51
  • \$\begingroup\$ I waited until I got my latest tests back from the printers.. perfect :) - now marked as the correct answer. \$\endgroup\$
    – Tetsujin
    Jun 29, 2017 at 9:47

Only perceptual rendering will take out of gamut colors and move the in gamut. The term Perceptual means gamut remapping to the closest color we humans perceive as a color match. (Metameric match)

Absolute Colorimetric Rendering intent does not map source RGB profiles white point to the media white point in the destination profile and produces a tint of color in some cases when simulating print on colored media. Any out of gamut colors from the source profile are clipped to the nearest in gamut color in the destination profile.

Relative Colorimetric Rendering intent maps the source RGB profiles white point to the media white point in the destination profile. Basically making the white points relative and "adapted white". Any out of gamut colors from the source profile are clipped to the nearest in gamut color in the destination profile.

If you don't have out of gamut colors in your image, then don't use a large RGB color space to render into. Convert using relative colorimetric rendering will be the most accurate.

Perceptual rendering is profile dependent. So depending on the manufacturer of the profile your results may vary. (buyer beware)


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.