Home Products Download Order Contacts


Subject: range tree usable for ray tracing?


I was recently asked by a coworker why I implemented our raytracer using
a kd-tree instead of a range tree as she was told that range trees are
much faster. My motivation was simple: I just had a look at Havran's and
Wald's PhD theses and followed them. One reason against range trees
would be the increased memory usage. However, our scenes are not that
large (< 10^6 triangles), so the additional factor log N might be
tolerable. Now I wonder whether range trees could actually accelerate
the ray tracing process since ray intersections are quite different from
range queries. And even if there's a theoretical advantage, the constant
factors might be significant for smaller scenes. Is there an obvious
point I missed? Is anyone aware of research in that direction?



View All Messages in comp.graphics.algorithms



Copyright 2006 WatermarkFactory.com. All Rights Reserved.