[argyllcms] Re: build rgb equivalent of cmyk profile

  • From: Graeme Gill <graeme@xxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Fri, 11 Dec 2015 09:37:02 +1100

Andrzej Andrez wrote:


I try to build rgb profile that is equivalent to other cmyk profile. I
found that fakeCMY and excel do the trick (I found comment TTBD i fakecmy
source code and I tweaked resulting ti3 to be rgb).

But I try understand
what fakeCMY is doing. It is possible to recreate effect of fakeCMY
manually with other tools like icclu/xicclu or fakeread? Source code in C
is too coplex for me and I can't figure it out.

I would guess you can make up values using other tools, although there
would be some maths involved. The main task is 1) to create a reasonable
3D CMYK "surface" within the 4D device space that gives you the full
gamut, and 2) drawing a correspondence between those CMYK value and
the CMY/RGB values you want in your output point set.

xicclu -fif gives you the ability to lookup a given CIE target value
with a given black generation curve, the trick is to only lookup
CMYK values that are going to be in gamut.

The way fakeCMY works is to first generate a CMY and neutral axis
set of points, and then figure out a scaling factor to stretch the CIE
values at black end of the CMY gamut to fully fill the CMYK gamut. It does
this by computing a mapping vector between CMY and CMYK black, and then
applying a scaled-by-distance-to-white version of that vector to
each point. The neutral axis points are treated similarly
but slightly differently to maintain their neutrality.

The result relies on collink interpolating most of the internals of
the colorspace, but in the intended application (as a CMY->CMYK
separation) this doesn't matter, because it all gets profiled,
so resulting CMY profile takes the separation into account.

Is it possible to make gamut file (and visualisation) of n-color profile
with argylcms?

Yes, although it's a bit limited. You would need to make an mpp profile
of the n-color .ti3 data, and then use mpplu to create a gamut surface from it.

I checked iccgamut and it is not working with n-color


It is a feature or bug?

It's a limitation of the current code.

Maybe someone knows other tool (open
source or free) to visualise gamuts including n-color profiles?

I don't know of any available tools, but then I haven't looked.

Graeme Gill.

Other related posts: