Path: utzoo!utgpu!utstat!jarvis.csri.toronto.edu!mailrus!tut.cis.ohio-state.edu!ucbvax!agate!shelby!polya!rokicki
From: rokicki@polya.Stanford.EDU (Tomas G. Rokicki)
Newsgroups: comp.graphics
Subject: Re: Color Quantization (was Re: Discussion of Computer Pornography)
Message-ID: <9445@polya.Stanford.EDU>
Date: 24 May 89 21:08:47 GMT
References: <4706@uoregon.uoregon.edu> <310@celit.UUCP> <8478@venera.isi.edu>
Sender: Tomas G. Rokicki <rokicki@polya.Stanford.EDU>
Organization: Stanford University
Lines: 16

raveling@venera.isi.edu (Paul Raveling) writes:
> 	Another technique that appears to work well is to weight
> 	selection based on pixel intensity. . . .

Wouldn't it make sense to do color selection in the HSV cone rather
than in the RGB cube, since the `distances' in HSV are more relevant
to perceived differences in color?  This shouldn't introduce too much
additional complexity (a map-to at the start and a map-from at the
end, but with some knowledge of what `points' in the cone are
permissible in the middle) but should yield better results.

Please, someone tell me I don't know what I'm talking about if this is
the case, but RMS error only makes sense if the error terms are
comparable, and this is certainly not the case in the RGB cube.

-tom