Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!eos!shelby!polya!rokicki
From: rokicki@polya.Stanford.EDU (Tomas G. Rokicki)
Newsgroups: comp.graphics
Subject: Re: How to map 24-bit RGB to best EGA/VGA palette?
Keywords: RGB EGA VGA color
Message-ID: <11540@polya.Stanford.EDU>
Date: 25 Aug 89 20:50:13 GMT
References: <3129@cbnewsm.ATT.COM> <7743@cbmvax.UUCP> <13319@well.UUCP> <4862@eos.UUCP>
Sender: Tomas G. Rokicki <rokicki@Polya.Stanford.EDU>
Distribution: na
Organization: Stanford University
Lines: 21

> What I've done is just dither r, g, and b independently and then
> combine them into a 3 (or more) bit image.  There are some interesting
> questions regarding the choice of the three matrices.

In my experiments, I've found that using identical matrices is better
than using different matrices, whether for ordered dither or classic
dither.  (Not much you can do about F-S.)  This has two advantages
and one disadvantage.

With coincident matrices, the output simply looks better; there are no
extraneous pixels of a very wrong color, just some that are almost off.
(Ie, if you had 0.8 r, 0.8 g, and 0.2 b, with coincident matrices, you
would never have a blue-only pixel, but with other matrices, you would.)

Secondly, at least on some printers, the ability to maximize use of black
ink seems to add contrast.

As a disadvantage, well, don't do color separations this way---it's
unlikely that your printer can duplicate the alignment that closely.

-tom