fisherm@infonode.ingr.com (Matthew Fisher) (06/29/91)
Hello All,
I recently read the article in Siggraph 1989 'A Ray Tracing Algorithm
for Progressive Radiosity' by Wallace and friends. I have an understanding
of the progressive hemi-cube approach and of ray tracing. I'm having a few
problems with ray tracing progressive radiosity though.
In the following situation, I don't see how the ray tracing progressive
aprroach will get the correct solution.
Suppose our world consists of the three square
polygons. One, L,is the light source and is far enough from the other two to
act like at point source. The polygon L faces the other two. The other two
polygons are of diffent sizes one is larger than the other, call it B.
The smaller is named S.
Now suppose all three are centered on the Z axis. L is at 100, B is at 0
and S is at 5.
Considering this situation from a ray trace perspective, both S and B will
be lit. A sharp shadow cast by S will be in the middle of B.
Considering the situation from a hemi-cube perspective. If the hemi-cube
is about L, all of S will recieve light. Most of B will also recieve light,
only the part blocked by S will not. Again the results are correct.
Now comes ray tracing progressive radiosity. The big change is that
only the vertice's form factors will be caculated. Suppose L
is the source. Suppose a corner vertex of B is the receiver. Rays will
be cast from B to various parts of L. If S is small enough or L is far
enough away, none of the rays cast from B to L will strike S. How then
are the proper shadows going to be placed on B?
I already have written a ray tracer. I would like to use it to do
progressive radiosity. I want to have a full understanding of the algorithm
before I start though.
Any commenta will be welcome. Please post answers to comp.graphics, my
mail doesn't seem to come through all the time. Here is my address if you
want to talk one on one.
fisherm@infonode.ingr.com
Take it easy
Matthew E Fisher