Path: utzoo!dptcdc!jarvis.csri.toronto.edu!mailrus!cornell!rochester!rocksanne!bozo!kirby
From: kirby@bozo.wbst128.xerox.com (Mike Kirby (co-op))
Newsgroups: comp.graphics
Subject: Re: "Different" 8-bit color to 1-bit converter
Message-ID: <721@rocksanne.UUCP>
Date: 17 Apr 89 14:27:13 GMT
References: <1296@rpi.edu>
Sender: news@rocksanne.UUCP
Reply-To: kirby@bozo.UUCP (Mike Kirby (co-op))
Organization: WRC, XEROX
Lines: 32

>I'm stuck on a monochrome sun 3/50 with no possibility of grey scale, and I
>was wondering if anyone out there knew of a program to convert an n-bit color
>or grey-scale image to 1 bit monochrome using a matrix of pixels to represent
>an individual pixel in the original file.
>I have access to some 8 bit to 1 bit conversion programs, but quite often,
>the picture is a 320x200 image, and it is sooo small on my screen, it's almost
>completely useless.  
>I've seen a program to do something like this before, but I can't remember
>where, and I know it wasn't on a sun.
>
>For example, there might be a 3x3 pixel array used to represent up to 10 shades
>of grey...

What it seems that you want to do is either halftoning, or some kind of
error diffusion.  I would recommend something like a floyd-steinberg error
diffusion. (you can get the article out of proceedings of the  S.I.D, vol 17/2
1976)  

Basically, this algorithm takes a pixel, determines from it's desired
intensity what the actual intensity would be.  For example, and intensity
of 1/4 (as shown by the article) would be black (intensity 0) on a monochrome
monitor.  Thus, we have an error of 1/4.  The idea is to distribute this
error to neighboring pixels in hopes that the errors will cancle out as we
go along.

This wasn't a very good explaination.  Check out the article for a better
description of the process.  It is a good technique.

Michael Kirby
mpk9172@ritvax.bitnet		(this works)
mpk9172%ritcv@cs.rit.edu	(flaky...but with a good tail wind I'll get it)
kirby.wbst128@xerox.com		(never tried...)