[argyllcms] Re: Gamut mapping problems in 1.1.0

  • From: Idea Digital Imaging <qcore@xxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 9 Nov 2009 15:14:59 +0000

On 9 Nov 2009, at 04:28, Graeme Gill wrote:

qcore@xxxxxxxxxxxxxxxx wrote:
The gamut mapping appeared to me to be working in that the separations themselves were as I expect (and love) from using Argyll -- smooth, detailed, minimised clipping etc -- but mapping to an oddly restricted version of the ISO Coated v2 gamut.

I need a specific example to track this down, as I haven't seen
anything like this in my testing.

That's probably because you've adjusted it so that it works exactly as it should do whereas I'd like it to work in a way that'd suit the tastes of London fashion photographers and retouchers :-)

I've grabbed part of the Bill Atkinson Lab Test image and converted it to Adobe RGB (1998):


Made a device link using gamut mapping in both 1.0.4 and 1.1 using collink -v -ip -G -cmt -dpp AdobeRGB1998.icc ISOcoated_v2_eci.icc

I converted the image using cctiff and then masked the results so that you can see the 1.0.4 version on the left and 1.1 on the right as strips:


Note that 255 255 255 --> 0 0 0 1.38 in the 1.0.4 conversion.

With real images the results when proofed/printed (I ran some FOGRA proofs just now) are subtle, but saturated originals like the sunset and the tree above it (on Bill Atkinson's test page) lose the visual impact that the non-gamut mapped versions have and that the 1.0.4 version does a good job at preserving:



(I bet Klaus prefers the 1.1 red leaves but I think they are a little dull...)

More delicate images like the rose on the test image also lose out as the "glow" of the petals is reduced by the 1.1 gamut mapping.

I tried both -ims and -is intents in 1.1 but still couldn't get the separations to match the saturated colours I'd been getting in 1.0.4.

Martin Orpen
Idea Digital Imaging Ltd

Other related posts: