Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site nyit.UUCP Path: utzoo!watmath!clyde!bonnie!akgua!whuxlm!harpo!decvax!linus!philabs!sbcs!nyit!ph From: ph@nyit.UUCP (Paul Heckbert) Newsgroups: net.graphics Subject: Re: Allocation of color map Message-ID: <135@nyit.UUCP> Date: Mon, 24-Jun-85 21:12:03 EDT Article-I.D.: nyit.135 Posted: Mon Jun 24 21:12:03 1985 Date-Received: Thu, 27-Jun-85 08:21:16 EDT References: <12800001@umn-cs.UUCP> Organization: NYIT Computer Graphics Lab., Old Westbury, N.Y. Lines: 37 > ... The problem is allocating the color > lookup map (8 bits deep) properly to match whatever distribution of colors > result from any single run [of my ray tracer] ... Thanks to Kurt Papke and Leo Hourvitz for recommending my SIG'82 paper. It sounds like Scott basically has two choices: (a) Generate and store the picture in full color (24 bits) and then quantize down to 8 bits as a post-process, by looking at the statistical distribution of the picture's colors. (b) Pick a colormap a priori and quantize all pictures to it. 3 bits red, 3 bits green, 2 bits blue works well. The first method is more difficult but can yield superior results. The latter method can be done on the fly, so that the picture is written to the frame buffer as it is computed, and it also has the advantage that all of the pictures will use the same colormap, which makes compositing easier. In either case I recommend dithering. [ADVERTISEMENT] If you're into other 8-bit frame buffer tricks, perhaps you'd be interested in another paper I wrote: Techniques for Real-Time Frame Buffer Animation FX'84, London, Oct. 1984 I'll mail you a copy if you send your USmail address. Paul Heckbert NYIT Computer Graphics Lab ucbvax!decvax!philabs!sbcs!nyit!ph allegra!sbcs!nyit!ph (Sorry about the delay on this response; we've had some network problems here for the past few weeks) -- ucbvax!decvax!philabs!sbcs!nyit!ph Paul Heckbert allegra!sbcs!nyit!ph NYIT Computer Graphics Lab