From: Darren Grant <dgrant@ke...>  20080527 21:00:34

Here's an interesting problem: How do you get a fast approximation of the largest oriented box that can fit inside a convex polytope? I've got a physics engine that really likes oriented boxes, and a convex volume that can shatter into hundreds of long flat fragments (think glass or crystal). Scaling down bounding boxes by a constant is already working OK, but it is quite a bit tougher for the simulation to resolve than if the boxes do not overlap initially. Adaptively growing a box from a center point seems to be the way to go, but are there any other promising approaches that I'm missing? Thanks, Darren 