Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!lll-crg!nike!ll-xn!adelie!axiom!gts
From: gts@axiom.UUCP (Guy Schafer)
Newsgroups: net.graphics
Subject: How is anti-aliasing done?
Message-ID: <323@axiom.UUCP>
Date: Mon, 15-Sep-86 17:00:09 EDT
Article-I.D.: axiom.323
Posted: Mon Sep 15 17:00:09 1986
Date-Received: Tue, 16-Sep-86 05:52:48 EDT
Organization: Axiom Technology, Newton MA
Lines: 29
Keywords: Perhaps dumb questions

This might be a bunch of dumb questions, for I am but a simple man.
How is anti-aliasing done?

I've seen the effects of anti-aliasing and am suitably impressed.  However,
I can't seem to think of a good algorithm to do it.  Just averaging the
(color, intensity, brightness?) of the two cells adjacent to any object
boundary sounds good but there are a few problems:

-Should you take the average of all eight cells adjacent to the cell to
be changed?  If not, along which axis is the average taken?

-The old value of a cell (before anti-aliasing) must be remembered or the
averages will be screwed up on any adjacent cells.  This seems inefficient
at best, stupid at worst.  (Is it me?)

-How do you determine the location of a boundary?  Any sufficient (color,
intensity, brightness?) difference between adjacent cells?  Or is it known
at the time the object is 'built?'

I'm not afraid of coding if I know the algorithm and am willing to play
around with the code to get good results, but obviously I'm a 'duppa'
(fog-bound child).  Does anyone know a common or good algorithm or have
some code hanging around (any language) that will give me a glimmer?

Thanks in advance.

	>< ...{ decvax!linus | seismo!harvard }!axiom!gts

	"Gilt are the chains of god."