Thread View: comp.graphics.research
3 messages
3 total messages
Started by Tom_Lane
Thu, 07 Jan 1993 19:17
Fast dithering techniques?
Author: Tom_Lane
Date: Thu, 07 Jan 1993 19:17
Date: Thu, 07 Jan 1993 19:17
18 lines
1248 bytes
1248 bytes
The free JPEG image compression software currently uses traditional Floyd-Steinberg dithering when reducing the number of colors in its output. Even with careful coding, F-S dithering is a remarkably expensive proposition... in our current release, dithering *alone* (not counting color selection) takes nearly as much time as JPEG decompression proper. If anyone is aware of cheaper dithering techniques that give results as good as F-S, or of implementation tricks to speed up F-S, I'd appreciate hearing about them. thanks, tom lane organizer, Independent JPEG Group tgl@cs.cmu.edu -- Moderated by SCRI Vis <> Submissions to: graphics@scri1.scri.fsu.edu Guy, John R. Murray <> Administrivia to: graphics-request@scri1.scri.fsu.edu
Re: Fast dithering techniques?
Author: Torben AEgidius Mogensen
Date: Sat, 09 Jan 1993 02:17
Date: Sat, 09 Jan 1993 02:17
30 lines
1968 bytes
1968 bytes
Tom_Lane@G.GP.CS.CMU.EDU writes: >The free JPEG image compression software currently uses traditional >Floyd-Steinberg dithering when reducing the number of colors in its output. >Even with careful coding, F-S dithering is a remarkably expensive >proposition... in our current release, dithering *alone* (not counting color >selection) takes nearly as much time as JPEG decompression proper. >If anyone is aware of cheaper dithering techniques that give results as good >as F-S, or of implementation tricks to speed up F-S, I'd appreciate hearing >about them. I once made a variant of F-S that was somewhat faster: instead of distributing the error to four neighbouring pixels, distribute it to only one. The idea is to make the choice of which pixel to distribute to random: in the simplest case you randomly choose either the next pixel on the same scan-line or the same pixel on the next scan-line. You need a fast way to generate random bits, bit a simple shift-register method works reasonably well. If you want slightly better quality, you randomly choose one of the four pixels that F-S would distribute the error to. The quality is not quite up to the standard of F-S, but nearly so. Torben Mogensen (torbenm@diku.dk) -- Moderated by SCRI Vis <> Submissions to: graphics@scri1.scri.fsu.edu Guy, John R. Murray <> Administrivia to: graphics-request@scri1.scri.fsu.edu
Re: Fast dithering techniques?
Author: Spencer W. Thomas
Date: Mon, 11 Jan 1993 23:16
Date: Mon, 11 Jan 1993 23:16
15 lines
1356 bytes
1356 bytes
In article <11660@sun13.scri.fsu.edu> torbenm@diku.dk (Torben AEgidius Mogensen) writes: > I once made a variant of F-S that was somewhat faster: instead of > distributing the error to four neighbouring pixels, distribute it to > only one. See Ulichney (MIT Press, 1988) for several variants of this scheme. -- =Spencer W. Thomas | Info Tech and Networking, B1911 CFOB, 0704 "Genome Informatician" | Univ of Michigan, Ann Arbor, MI 48109 Spencer.W.Thomas@med.umich.edu | 313-747-2778, FAX 313-764-4133 -- Moderated by SCRI Vis <> Submissions to: graphics@scri1.scri.fsu.edu Guy, John R. Murray <> Administrivia to: graphics-request@scri1.scri.fsu.edu
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads