On Jun 22, 2004, at 9:40 AM, Andre Schützenhofer wrote:
Roughly, mapping colors onto the gamut boundary
normally preserves
L-Values which often preserves differentiation (You might imagine an
LAB-image completely desaturated (which will appear gray) - all that
happend was moving all colors towards L-axis. Of course you will lose
color but not necessarily any detail.) In addition, most
profile-making programs add some 'perceptual treatment' to
out-of-gamut-colors within colorimetric intent - instead of simply
moving linearly onto the gamut boundary.
Yes, my explanation was a rough description. If you graph the effects
of different rendering intents with ColorThink, for example, you can
see this. The whole gamut is affected in every intent, and the line
between Perceptual and Rel Col is more blurry than one would think. But
to a first approximation, the effects are in accord with the simplified
scheme of Perceptual = in-gamut scaling and Rel Col = in-gamut
unaffected, out-of-gamut pulled into gamut boundary.
It's interesting to graph similar profiles made with different vendors'
software. You can clearly see that there is no single method of
implementing a rendering intent.
Regards,
Mike Cummings
CHROMiX Technical Support
1-866-CHROMiX (247-6649)
cummings(a)chromix.com