From albanycs!leah:rsb584 Mon Jan 18 16:56:49 1988
Received: by albanycs.albany.edu (5.54/4.8)
	id AA03303; Mon, 18 Jan 88 16:05:07 EST
Date: Mon, 18 Jan 88 16:05:04 EST
From: albanycs!leah:rsb584 ( Raymond S Brand)
Received: by leah.Albany.EDU (5.58/1.1)
	id AA17810; Mon, 18 Jan 88 16:05:04 EST
Message-Id: <8801182105.AA17810@leah.Albany.EDU>
To: albanycs:beowulf!rsbx
Subject: dither7.txt

>From lou@aramis.rutgers.edu Sun Jan 10 23:46:33 1988
Path: leah!uwmcsd1!ig!jade!ucbcad!ames!rutgers!aramis.rutgers.edu!lou
From: lou@aramis.rutgers.edu (Lou Steinberg)
Newsgroups: comp.graphics
Subject: Re: Advanced Dither Needed
Keywords: dither
Message-ID: <2708@aramis.rutgers.edu>
Date: 11 Jan 88 04:46:33 GMT
References: <3703@ames.arpa> <2741@masscomp.UUCP>
Distribution: na
Organization: Rutgers Univ., New Brunswick, N.J.
Lines: 20

In article <2741@masscomp.UUCP> garyo@masscomp.UUCP (Gary Oberbrunner) writes:

> [for] Floyd & Steinberg's dithering algorithm [...]
> On some images it can help to add some random noise; for instance if you
> have a smooth (flat-shaded) surface that's just below one of the colors in
> the map, the error will build up slowly until you get a line of brighter
> color somewhere in the surface.  A bit o' noise gets rid of that line nicely.

The other way to get rid of it is to alternate scanning left to right
and right to left, i.e. even scan lines one way, odd scan lines the
other.  The line of brighter color essentially comes because pixels on
that line do not communicate - there is no path for the error value at
one to affect the value chosen for the other.  By scanning in a zig
zag fashion, you ensure that each pixel communicates with every other
pixel scanned after it.
-- 
					Lou Steinberg

uucp:   {pretty much any major site}!rutgers!aramis.rutgers.edu!lou 
arpa:   lou@aramis.rutgers.edu