Marti wrote:
Anyway, some answers. This particular profile is a "mix" between v4 and v2. If you list the tags, you will find there is a chromatic adaptation tag present. Well, lcms at its 1.14 incarnation implements absolute intent as "no adaptation", so, it uses the chromatic adaptation tag to "undo" the chromatic adaptation and therefore, recover the true primaries. Next version will have a way to control if absolute intent should behave as observer is fully adapted (this is called the ICC-absolute intent) or as observer not adapted at all (very useful in a match to screen environments).
Right. After reading the first couple of chapters of the ICCV4.2 spec., I see where this at. Yes it appears that the only way to provide V4 compatibility and backward compatibility is to create a new pseudo-intent (something like "icICCAbsoluteColorimetric"), as well as the backward compatible "icAbsoluteColorimetric", or "icRealAbsoluteColorimetric". This whole change is a real puzzle. It's wrong in so many ways, breaking backwards compatibility with with V2 profiles and CMM's, and not addressing real problems like the continued use of the "wrong von-Kries" absolute to relative transform. (Reminds me rather unfortunately of the change to the L*a*b* PCS encoding in V4.) It all appears to have been driven simply by some users surprise at getting "blue" output when they link CRT profiles to print profiles using absolute intent. A simple example shows what's wrong with it: I have a display with a white point of D65. When I profile it according to ICCV4, I end up with a white point of D50, and colorimetric data that has been Bradford transformed from D65 to D50. If I lookup the media color using relative colorimetric I get D50. If I lookup the media color using ICCabsolute I get D50. If I look at the screen next to a (white) piece of paper illuminated with a D50 light, the display looks blue. If I look at the display on it's own, it looks white. I have some blue tinted paper I'm using in a printer, and I do a printer profile for it using a colorimeter that returns D50 illuminant XYZ values. When I profile it according to ICCV4 I end up with a white point that is blue (reflecting the tint of the paper, say around D65 white point), and colorimetric data that has been "wrong von-Kries" transformed from D65 to D50. If I lookup the media color using relative colorimetric I get D50. If I lookup the media color using ICCabsolute I get D65. If I look at the blue tinted paper next to a white piece of paper illuminated with a D50 light, the blue tinted paper looks blue. If I look at the blue tinted paper on it's own, it looks white. In other words, for identical visual situations, the two profiles give different ICCabsolute results. Three examples of the results: If I link my blue paper to a normal white paper output profile using ICCabsolute, I would get "blue" output (the very complaint that seems to have triggered this change when applied to display profiles, although it is all working as you would hope absolute to work). If I link my display to my blue paper output profile using ICCabsolute, I would get "yellow" output (This change has "fixed" things in one direction only). If I try and reproduce spot colors on the display using ICCabsolute, hoping to be able to check for an exact match (as supposedly ICCabsolute is intended), then they won't match - the display output will be too blue. Graeme Gill. ------------------------------------------------------- SF.Net email is sponsored by: Discover Easy Linux Migration Strategies from IBM. Find simple to follow Roadmaps, straightforward articles, informative Webcasts and more! Get everything you need to get up to speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click _______________________________________________ Lcms-user mailing list Lcms-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/lcms-user
