[argyllcms] Re: fwa compensation unexpected results
- From: Graeme Gill <graeme@xxxxxxxxxxxxx>
- To: argyllcms@xxxxxxxxxxxxx
- Date: Fri, 05 May 2006 23:36:40 +1000
Roberto Michelena wrote:
I agree that modelling the spectra is necessary to have a hope of
modelling perception. But that's not all, evidently, because I don't
see that paper as blue.
And I wouldn't normally expect you to see it as blue. You will
adapt to seeing it as white. That doesn't make the B* = -13
value invalid - that value is for the white of the paper
relative to the fixed reference point of D50. It's the
value you would get if you pointed a spectral teleradiometer
at the paper when illuminated by D50.
Case in point: my reference is Fogra or Gracol v7, with paper white
about 95,0,-2; moderate (usual) amount of FWA in the offset coated
paper is giving you that -2, while in fact we see it as perfectly
white (seen individually).
Of course.
My proofing paper is b=-8.8 with L=95 (lots of FWA), which also if
seen individually looks white.
If we put both papers side by side, suddenly the offset paper looks
yellowish. Not that the proofing paper looks bluish, it's the other
way around. Perception here is the result of adaptation.
As expected.
We make an abscol proof, and the proofer sees b=-2 reference, b=-8.8
proof, applies a lot of yellow ink. Put both the proof and the
reference together, and you see the proof is much yellower. Of couse,
the system 'saw' the proof paper as quite blue when we saw it as
white, in fact 'whiter' than the reference offset paper.
As it should. If it's all working, the papers should now match (although,
maybe not perfectly, since adding the yellow will pull the L down.
If the L goes down too much, it's not a good match, and your reference
is out of gamut as far as the proofing paper is concerned.)
Of course if you've made your measurements under the 'A' illuminant
of the instrument, and you're actually viewing under an illuminant
which is much poorer in UV (relative to other wavelengths) than 'A'
illuminant, then the proofing paper is not actually b -6.8 bluer
than the reference under those condition, and too much yellow will
have been applied.
So we need to align the system with what we perceive. We then use a UV
filter to measure the proof characterization chart, resulting in b=-3
in white. Now the system applies just a tad of yellow to compensate,
we put both together, and they match.
They should be way out if you're viewing under D50. Your proof would be
expected to be about delta b = -5, too blue. If it's not the case, then
either your profiling system is broken, or you're not viewing under D50.
If they do match quite well and your profiling system is not broken,
then you're got a better match by fudging the instrument illuminant to
be closer to your actual viewing illuminant by switching in the UV filter.
In or Out, only two choices. Argyll's FWA compensation has more subtlety
than that.
When most commercial softwares talk about fwa compensation, they mean
simulating this effect. And this indeed improves the visual match in
proofing, as explained above.
And as I've explained, it may seem as simple as bumping the white
point in the +b direction, but that's not the right way way of fixing
things. If you're not prepared to try it, then there's no point in me
saying anything more on the subject, I think I've explained
the situation quite clearly enough.
Graeme Gill.
Other related posts:
