Microsoft PRT tools
Has anyone played with the Microsoft Direct X PRT engine code? When I run my mesh through ID3DXPRTEngine::RobustMeshRefine() or ID3DXPRTEngine::ComputeDirectLightingSHAdaptive() , the code seems to hang on some meshes that I give to it (ComputeDirectLightingSHAdaptive() always seems to stick at 50%). I've cleaned up the mesh and validated it, so I'm at a loss for what I need to do:
D3DXCleanMesh(D3DXCLEAN_OPTIMIZATION,theMesh,aAdjacencyIn,&theMesh,aAdjacencyOut,NULL);
D3DXValidMesh(theMesh,aAdjacencyOut,NULL); // seems to think my mesh is valid
Does anyone have any clues what I might be doing wrong?
Thanks!
I haven't played with it, but some of the PRT precomputing functions are supposed to be quite slow... like taking hours, or days, on complex meshes. That's about the extent of my knowledge on the subject. Try letting one of these models sit overnight, and if it's still going after that then consider it to possibly have crashed.
it shouldn't take long at... (days?!)
for 50k poly mesh, expect a few minuets (3 bounces)
what probably stalls your compile is if you've got the adaptive tesselation setting on... which will result in insane compile times for negligable quality increase (especially for per vertex... you should just author the mesh to work well with per-vertex PRT... not let the compile tesselate)
for 50k poly mesh, expect a few minuets (3 bounces)
what probably stalls your compile is if you've got the adaptive tesselation setting on... which will result in insane compile times for negligable quality increase (especially for per vertex... you should just author the mesh to work well with per-vertex PRT... not let the compile tesselate)
Quote:Original post by Anonymous Poster
it shouldn't take long at... (days?!)
for 50k poly mesh, expect a few minuets (3 bounces)
what probably stalls your compile is if you've got the adaptive tesselation setting on... which will result in insane compile times for negligable quality increase (especially for per vertex... you should just author the mesh to work well with per-vertex PRT... not let the compile tesselate)
Too high subdivision depth is what can potentially slow down the solver considerably, not the fact that adaptive tessellation is just enabled.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement