[argyllcms] Re: Confused with PS, softproofing and monitor profile intents

  • From: Vallo Kallaste <kalts@xxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Sun, 7 Nov 2010 00:02:36 +0200

On Sat, Nov 06, 2010 at 02:14:53PM +0100, Klaus Karcher
<lists@xxxxxxxxxxxxxxxxx> wrote:

> Do you mean the the Rendering Intent Flag in the profile header?
> "Note that this flag might not have any meaning until the profile is
> used in some context, e.g. in a DeviceLink or an embedded source
> profile." (Spec. ICC.1:1998-09)
> As long as there are no CLUTs in the profile, it doesn't matter
> whether it is set to relative or perceptual: the CMM will fall back
> on the TRC/matrix model in any case. Only absolute colorimetric
> might make a difference in an application that makes use of the
> Rendering Intent Flag, but such applications are really rare.

Ok then, the flag does not matter. It's weird that ICC specification
even allows for such a thing.

> >How are all the profiles and intents interacting? Is PS creating
> >device link profiles on the fly, considering also source image
> >gamut?
> Photoshop (or the CMM) creates a device link profile on the fly, but
> a very simple one: there is no dynamic gamut mapping.
> The softproof link always works like this
> image -> printer -> monitor
> The image -> printer transformation is affected by the rendering
> intent and black point compensation settings you set for the
> softproof
> and the printer -> monitor transformation is affected by the
> "simulate paper" and "simulate black" settings:
> - simulate paper and black means absolute colorimetric
> - simulate only black means relative colorimetric without BPC
> - simulate neither paper nor black means relative colorimetric with BPC
> It doesn't matter whether your monitor profile contains a real
> perceptual intent or not: Photoshop uses the colorimetric intents
> for softproofs in all cases (unless you hack the tag table of the
> monitor profile ;-)
> When you just display an image in Photoshop (without softproofing)
> the transformation image -> monitor is always rel with BPC.
> Perceptual intents in monitor profiles are therefore not of much use
> in Photoshop. The only way to use them are expicit manual
> tranformations (e.g. by converting an image to monitor space with RI
> perceptual)

Thank you for comprehensive answer and a hint how PS is ignoring the
RI's in monitor profile.
Are there any common apps in photo viewing/editing field which
actually let one use/choose the perceptual RI if it's present in the
monitor profile? I once asked why dark tones are clipped in PS and
Graeme replied:
The usual reason is the lack of appropriate gamut mapping.
If you connect a source profile with a zero black (ie. such
an idealized sRGB or AdobeRGB type profile) to a profile
that has a non-zero black point (ie. like a real world
diplay profile) without gamut mapping, the blacks get clipped.

[For colorimetric they always get clipped, because that's what
you've asked for in selecting Colorimeric.]

So generating cLUT profile with different RI's doesn't help at all.
I must honestly say that it is not a problem anymore, simple matrix
monitor profiles with sRGB TRC do not clip darks anymore. At least
compared to similar matrix profiles generated by Xrite software.
Something must have been changed (for good) since I wrote about it
and the solution to this problem was not in gamut mapping.

On Sat, Nov 06, 2010 at 04:22:59PM +0100, Claas Bickeböller
<lists@xxxxxxxxxxxxxxxxx> wrote:

> Hi Vallo,
> in addition to Klaus' answer.
> Maybe the lecture of our Softproofing Handbook gives additional valuable 
> input.
> http://forschung.fogra.org/index.php?menuid=62&reporeid=122&getlang=en
This is a fine piece of distilled information about color
management basics, thanks for making it available!

Other related posts: