Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!usc!wuarchive!sdd.hp.com!hplabs!hpfcso!hpfcdc!koren
From: koren@hpfcdc.HP.COM (Steve Koren)
Newsgroups: comp.sys.amiga
Subject: Re: Imagine's CSG Spheres
Message-ID: <5500044@hpfcdc.HP.COM>
Date: 4 Jan 91 04:54:43 GMT
References: <2062@beguine.UUCP>
Organization: HP Fort Collins, Co.
Lines: 34


[I appologize if this response appears more than once; the notes
 poster program burped repeatedly ]

Viet Ho writes:

> I did some quick rendering tests on Imagine today and noticed
> the edges of the CSG spheres are *FACETED*  Now I've double checked
> to make sure these are CSG spheres in the Stage editor.  There's
> only one sphere in the scene.   Is Imagine doing polygon mesh
> approximation on their CSG spheres for the render process?  
> Why would they do such a foolish thing?   

There is a common technique called "Gereaud shading" (yeah, I know I
badly mangled the spelling :-) ).  It involves representing curved
surfaces as a set of polygons, and making the curved surface look
smooth by interpolating the normal vectors between centers of the
polygons.  Since the way the light "reflects" is determined largely
by the normal vector, the surface will look smooth even though it
is composed of polygons.  Of course, you will still see artifacts of
this near the edges of the objects.  If you look closely at some
of the "computer animation" video tapes floating around from the
state of the art animation houses (ILM, etc), you can often see
this technique being used.

Now, I haven't seen Imagine, so I don't know that is necessary what
they are doing. But it could be.  It has several advantages: firstly,
it is easier to render polygons than curved surfaces, and secondly,
arbitrary curved surfaces can be modelled easily that way, where
trying to model them using some other method would be more difficult.

Hope this answers your question.

  - steve