Is there a reliable way to interpolate a large point cloud volume when running on a parallel MPI server?
Usually to interpolate a point cloud, I use Point Volume Interpolator. It works well in serial, but it is greyed-out when running on a parallel server. Checking the servermanager xml, it looks like the proxy has multiprocess_support="single_process" set. So, that doesn’t work.
Gaussian Resampling is a fallback. It technically runs, but the result is pretty unusable. There are a lot of artifacts because it does not manage boundaries. It also throws warnings and volume renders wrong. I think this is because the filter is not respecting extent requests.
I guess Delaunay 3D is a fallback. It works OK on smaller data, but it is known to be really problematic for larger data. It also introduces its own artifacts because it does not respect processor boundaries.
Is there a 4th option that I am missing?