skinner@saber.UUCP (04/21/86)
> <I really dislike inclusions> > > In article <396@cit-vax.Caltech.Edu> jon@cit-vax.UUCP (Jonathan P. Leech) writes: > >In article <1664@sdcsvax.UUCP> hutch@sdcsvax.UUCP (Jim Hutchison) writes: > >>... By using an ordered color dither, I get a reasonable image, and > >>maintain a static map configuration. > >>... > >>Any suggestions? What I have now is very pleasantly displaying the 24bit > >>images we have received from USC (mandrel, milk drop, faces, peppers, park). > > > > You might try Floyd-Steinberg dither (I believe this is also known > >as 'minmized average error' dither. It works basically like this: > >... > > My implementation of this is ~ 3x as slow as ordered dither on > >Suns and HP bobcats - enough to be quite annoying. > >-- > I also implemented the Floyd-Steinberg dither, first in true floating point (quite slow on a workstation), then in integer (not much faster, due to extra precision to prevent truncation errors). Then, I finally realized that the quantization value and the error terms could be pre-computed and stored in a look-up table. The resulting program is about 4x as fast as the floating point version. If anyone is interested, I will post the code. It has some limitations, such as dithering to a destination of at most 8 bits, but the 8 can be any combination of red, green, or blue bits. It also handles less than 8 bits well, as would be needed with an 8 color printer. Of course it still exibits the problems of F-S dither. ------------------------------------------------------------------------------ "You know something people? I'm not black, But there's a whole lots of times I wish I could say I'm not white" Name: Robert Skinner Snail: Saber Technology, 2381 Bering Drive, San Jose, California 95131 AT&T: (408) 435-8600 UUCP: ...{decvax,ucbvax}!decwrl!saber!skinner ...{amd,ihnp4,ittvax}!saber!skinner