[argyllcms] Re: Soft-proofing in Photoshop with Argyll - correction to my last post!

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Thu, 12 Jun 2014 10:44:59 +1000

robert@xxxxxxxxxxxxxxxxxx wrote:


> I use Lightroom to print and Photoshop CS6 to soft-proof.  Both use the
> Adobe ACE CMM engine.
> My document's profile is Adobe RGB.
> The monitor profile is Eizo_CG277_Native_6500K_80cd.icc
> Photoshop's working space is Adobe RGB.
> I print the document from Lightroom using Relative Colorimetric with the
> HP_Z3100_Canson_Platine.icc profile (produced by colprof without -i & -f).
> I softproof in Photoshop.
> I set the soft-proof "Device to Simulate" to either


Well, unfortunately I'm not familiar with any of the details of all those 
and certainly am not in a position to investigate or duplicate such a workflow,
so I can't really help with such detail.

I can explain in general how color profiles work and how CMM's typically
work with them, but it's up to you or others familiar with the details
of particular systems to figure out how it is really working.

> I set the soft-proof "Rendering Intent" to Relative Colorimetric, "Black
> Point Compensation" on, "Simulate Paper Color" on.  

That sounds rather contradictory. "Simulate Paper Color" == absolute 
in my books, yet you've selected Relative Colorimetric. So it's hard to guess
what what Photoshop is actually doing.

It may well be doing a mixed intent transform which is making the input
print profile be absolute colorimetric and the output display profile
relative colorimetric. This is not true soft proofing, since there will
be a white point shift (unless the display is set to the same white point
as the print viewing conditions). Also, Black Point Compensation will
probably wreck any true soft proofing, since it will stretch the actual
paper black down to the display black.

But I gather many people like this sort of "print simulation" since
it makes for a memory match, and avoids the adaptation issues that arise
when other UI elements are on the screen, but it's unlikely to pass scrutiny
in a side by side proof comparison with a print.

> BTW, I don't know what Photoshop CS6 (or Lightroom) is doing, but all
> rendering intents simulate the paper color: so using the
> HP_Z3100_Canson_Platine.sp in all cases does not give the simulated paper
> white with Solux lighting (it gives the D50 simulated paper white), whereas
> using HP_Z3100_Canson_Platine_Solux_Proof or
> HP_Z3100_Canson_Platine_Solux_Proof_Not_FWA_Compensated does, again in all
> cases.  I've also created D50 and D65 profiles (-iD50 -fD50 and -iD65
> -fD65): when soft-proofed with the D50 profile changing from the print to
> the softproof profile makes no difference, as expected; selecting the D65
> profile with simulated paper color ON gives a much bluer color than with
> simulated paper color OFF, which seems wrong as the monitor was profiled to
> 6500K and currently has a CCT of 6600K).

Doing a true soft proof with a typical 65K white point display filled
with UI elements is not easy. You either have to be able to switch
the white point of all the elements on the display including the UI
(something that can only be done by a VideoLUT/vcgt change),
or do a full screen soft proof to remove all the UI elements, or
calibrate the screen to have the same white point as the paper you
are trying to proof. In any case, a true soft proof will use
absolute colorimetric rendering throughout (something ICC V4
profiles are incapable of by the way, although calibrating the screen
white point is a workaround).

In an ideal soft proof setup you should be able to point a tele-spectrometer
at the print in the lighting booth and the display and get the same XYZ values
for each picture element.

Graeme Gill.

Other related posts: