Xref: utzoo comp.graphics:9264 comp.windows.x:16744
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!uwm.edu!cs.utexas.edu!usc!venera.isi.edu!raveling
From: raveling@isi.edu (Paul Raveling)
Newsgroups: comp.graphics,comp.windows.x
Subject: Code for quantization algorithm
Message-ID: <11309@venera.isi.edu>
Date: 10 Jan 90 00:38:43 GMT
Sender: news@venera.isi.edu
Reply-To: raveling@isi.edu (Paul Raveling)
Organization: USC Information Sciences Institute
Lines: 30


	To answer what's quickly becoming a common request, C
	code for the new color quantization algorithm I described
	WILL be publicly available, but probably not until about
	the start of February.

	Before being unleashed, it needs some final detailing
	and more testing.  It'll be merged into an updated copy
	of the Img Software set that should also fix some minor
	bugs in the copy distributed with X11R4.


	Since about 50% of the email requests I've received asks
	about PC's, please note that one of those unfinished details
	is trying out 5-bit and 4-bit prequantization, instead of
	6-bit.  This could shrink a 1 megabyte color pointer array
	to 128K or 16K, respectively.  Another detail on PC's is
	that certain critical code uses if/then/else logic that
	flies on my 68K-based workstation, but it'll bog an 80x86
	down badly due to queue flushes.

	Performance loss due to queue flushes will probably be
	about a factor of 10 in critical per-pixel logic.  I'm sure
	not everyone with an 80x86 is ready to buy something better,
	so I'll (again) sweat some blood to minimize flushing.


----------------
Paul Raveling
Raveling@isi.edu