js9b+@andrew.cmu.edu (Jon C. Slenk) (05/19/91)
I just looked at a paper on something called vista ray-tracing. The paper
makes it look like this method is really nice. Basically, the method
undersamples, thus reducing the time required to render an image. Certain
methods are used to make the picture as good as possible: basically, the
important parts of the image are given a lot of time and the
not-so-important parts are just quickly dealt with. I believe that the way
the authors determine the important parts is to use data from the
perspective projection ("vista") of the image to be generated.
I wanted to ask a couple of things: Has anyone implemented this method or
read the paper? If so, what do they think of it? Email, please (if enough
people care to hear the responses I may get, I will summarize). I have a
friend who might have a shot at writing a nifty ray-tracer for a parallel
processing system, and I figured he might be able to get down to a second
a frame, or even better with this system.
TIA,
-Jon / js9b+@andrew.cmu.edu