Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!sri-spam!sri-unix!hplabs!hao!noao!arizona!megaron!rogerh
From: rogerh@megaron.UUCP
Newsgroups: net.graphics
Subject: Re: Possible way of anti-aliasing.
Message-ID: <1229@megaron.UUCP>
Date: Wed, 8-Oct-86 15:56:42 EDT
Article-I.D.: megaron.1229
Posted: Wed Oct  8 15:56:42 1986
Date-Received: Sat, 11-Oct-86 20:17:20 EDT
References: <280@joevax.UUCP> <265@hoqam.UUCP> <7706@sun.uucp> <109@pixar.UUCP> <1234@turtlevax.UUCP>
Reply-To: rogerh@arizona.edu (Roger Hayes)
Distribution: net
Organization: Dept of CS, U of Arizona, Tucson
Lines: 10

Ken Turkowski posts a way to analytically anti-alias circles (thanks!),
but claims that this shows that it's possible to anti-alias "curved
surfaces".  Unfortunately, flat circles are different from curved
surfaces.  To anti-alias the rendition of a curved surface properly,
one must anti-alias any sharp-edged color changes; for example, the
edge of a highlight.  To do this analytically requires an analytic (and
tractable) description of the projection of the colored area.  I think
that this is not, in general, computable -- think of a painted scene
reflected by a cubic patch.  It's as hard as differentiating the
luminance function.