Path: utzoo!utgpu!watmath!clyde!att!ucbvax!husc6!rutgers!okstate!mbc
From: mbc@a.cs.okstate.edu (Michael Carter)
Newsgroups: comp.graphics
Subject: Re: Raytracing speedups....
Message-ID: <4303@okstate.UUCP>
Date: 29 Dec 88 20:22:41 GMT
References: <7839@versatc.UUCP>
Organization: Oklahoma State Univ., Stillwater
Lines: 27

From article <7839@versatc.UUCP>, by ritter@versatc.UUCP (Jack Ritter):
> 
> A simple method for fast ray tracing has occurred to me,
> and I haven't seen it in the literature, particularly
> Procedural Elements for Computer Graphics.
> It works for primary rays only.
> 

     I tried something like this a while back in an early version of my
Hypercube Ray Tracer.  As you described, I placed a minimum sized box
around the image of the object as projected onto the viewing plane, and
tested primary rays against it as a first-cut culling scheme.  At that time,
the ray tracer was using the NAIVE object intersection method.  The best
improvement I ever got from this scheme was about 20% overall speedup for
scenes with small numbers of large objects.  It seemed that there were
just too many secondary rays being cast for the decreased number of primary
rays to have much effect on the overall time.  This is especially true in
scenes where large reflective or partially reflective objects exist -- their
shadow testing rays and reflected rays just overwhelm any savings gained in
casting the primary ray.
     This is admittedly not an extensive test, the there is room for further
experimentation, but the extra code incurred for such a modest gain was not
worth the trouble in my estimation.  Sorry.  Keep the ideas coming, though!

-- 
Michael B. Carter (mbc@a.cs.okstate.edu)
Department of Electrical and Computer Engineering, Oklahoma State University