Re: Absolute vs Relative Colorimetry in the ICC Profiles?
Re: Absolute vs Relative Colorimetry in the ICC Profiles?
- Subject: Re: Absolute vs Relative Colorimetry in the ICC Profiles?
- From: "Bruce J. Lindbloom" <email@hidden>
- Date: Fri, 21 Sep 2001 14:12:59 -0500
Roger Breton wrote:
>
Which colorimetry gets encoded in an ICC profile?
Relative to the PCS, which is D50, 2 degree observer.
>
The 1998-09 ICC Spec on p.82 exhorts that "The PCS is based on Relative
>
Colorimetry". Fine. It is not clear that profiles have to incorporate
>
relative colorimetry as well. I am just trying to find out what to expect in
>
a profile A2Bx and B2Ax tags: relative or absolute colorimetry?
Relative.
>
If, indeed, profiles are encoded "relatively" then this means that makers of
>
the ICC profile have to convert the absolute colorimetric measurements read
>
off the various characterization targets, by color measuring instruments, to
>
some relative form through a choice of Bradford or Von Kries methods, in
>
order to populate the LUTs (after interpolation)?
Yes.
>
Otherwise, if profiles are encoded "absolutely" then this would mean that
>
the makers of the profile just go ahead and populate the LUTs using the same
>
absolute colorimetric data (massaged for interpolation) that was read off
>
the original characterization target? Logical?
Not applicable -- doesn't work this way.
>
If this second scenario turns out to be true, would it be then mean that, at
>
run-time, the CMM makes the absolute data it finds in the ICC profiles
>
"relative" by chromatically adapting to the paper white, in both source and
>
destination profiles, using Bradford or Von Kries, pulling the information
>
from the MediaWhitePoint Tag, I understand thus making both sides of the
>
equation L = 100, a=b=0, so that C=M=Y=K=0 and R=G=B=255?
It works the other way around. Adaptation is performed *by the CMM* only
when absolute colorimetric rendering is requested. At that time, the CMM
attempts to reconstruct the original absolute colorimetry on-the-fly by
using the media white point to "un-adapt" the relative colorimetry. Since
the ICC profile format has no provision for the profile creator to indicate
the adaptation algorithm used in creating the profile, the CMM must
arbitrarily guess at a method, and therefore correct absolute colorimetry
*cannot* be guaranteed in the present ICC architecture.
>
Lastly, the ICC mandates on p.118 that the "coordinates of the output paper
>
be *adapted* to the PCS Illuminant"? Wouldn't that defeat the purpose of the
>
media white point data in the first place? Why wouldn't the Media White
>
Point be encoded in "absolute" terms?
The media white point is the D50, 2 degree observer measurement of the
paper. It is used by the CMM only for creating absolute colorimetry (i.e.
"un-adapting" as explained above).
--
Bruce J. Lindbloom, Pictographics Intl. Corp.