> Yes I know 'tis strange to be commenting on my own paper, but
> after only one comment _somebody_ had to get the ball rolling.
I guess I can make a few comments.
The list of references is a bit dependent on Graphics Gems. You might
also consider citing the source for the images on p27,28. I suppose they
might be original but given that the "devil's curve" surface occured as a
result of a transcription error (it was supposed to be a solid of
revolution), It seems unlikely that the same coefficients would be chosen
Given that all of the techniques described are present in existing
tracers, many with source available, you should probably reference any of
them you got information from. Ones that have almost all features you
describe, distribute source, and come to mind immediately are: POV-Ray,
Rayshade, RTrace, and VORT/ART.
The description of CSG on p68, while perhaps correct, is a terrible way to
do it in a raytracer. It's quite rare for that approach to beat finding
intersections from front to back and returning as soon as you find one
that isn't removed by the CSG. (Best to do it in conjunction with your
spatial partitioning routing.)