Re: [Lcms-user] XYZ values of colorants seem different?

[Graeme G. <gr...@ar...>]

ma...@li... wrote:

> Of course I had. Unfortunately with no results. The facts are, v4 has not 
> changed anything on the interpretation. It just happens v2 didn't specify 
> this behaviour. As a side effect, sRGB, AdobeRGB, CIERGB, AppleRGB, 
> ColorMatchRGB and all workspaces implemented as display profiles are 
> considered to be wrong.

I don't accept that. The spec. was ambiguous, but certain very widely
accepted profiles (like sRGB) and CMMs (like Adobe) interpreted it in
a way that worked perfectly well. Now V4 "clarifies" things (but not
really, given the contradictions about what absolute intent is meant to
be useful for), and makes the new profiles incompatible, reduces
functionality, and causes lots of problems.

> So, in order to get rid of all those issues, I've introduced the adaptation 
> state function. At first I intended to use a separate absolute colorimetric 
> intent, but this would create even more confusion to enduser, so, I think is 
> better to keep a single intent numbered '3' for absolute and provide an 
> additional function to control if the display should do a chromaticity match 
> or behave in this weird ICC-absolute way. Maybe in a future this function 
> would control incomplete states of adaptation. That would be good for 
> softproofing in mixed environments, but would need to use an appearance 
> model, so it is still rocket science right now.

Having thought it through, I would actually take the opposite approach.
Leave the CMM "absolute colorimetric" behaviour backwards compatible with
V2 (so that users and applications don't see any change between V2 and V4),
and add a new "ICCV4absoluteColorimetric" intent for those who prefer
the V4 behaviour.

Graeme Gill.