> *ALSO*, the simulated annealing approach runs more quickly...it completed in
> less than 60% of the time of the other one.
>

So after some more testing, I think the issue might actually be that the nearest
point algorithm using the kd-tree approach might be yielding suboptimal results,
which is requiring the isosurface code to do more samples, which is what is
actually slowing it down.

Basically, each sample is faster, but it's doing a lot more samples.

Either there is a bug in my kd-tree implementation, or there is something
fundamentally incorrect with the approach. Will have to do some more testing to
narrow down which it is.

It also occurs to me that I should be able to *somehow* use the mesh's bounding
box tree to implement a fast closest point query. Or maybe a parallel bounding
sphere tree?

Post a reply to this message