on which aws instance meshlib sdk can perform well

[dhanraj-khatal]

I am using c5n.2xlarge instance to run the meshlib python sdk API. to clean up the mesh and rebuild the mesh it is taking too much time. can you suggest any particular configuration in which meshlib can perform well?

[Grantim]

Hello!

Performance depends in multiple parameters, could you please tell us more about your cleanup workflow and show example of input data?

[dhanraj-khatal]

hi @Grantim ,
i am using the bellow method to cleanup the mesh

def heal( mesh : mm.Mesh, voxelSize : float, decimate : bool = True )->mm.Mesh:
	numHoles = mm.findRightBoundary( mesh.topology ).size()
	oParams = mm.GeneralOffsetParameters()
	if (numHoles != 0):
		oParams.signDetectionMode = mm.SignDetectionMode.HoleWindingRule
	oParams.voxelSize = voxelSize
	resMesh = mm.generalOffsetMesh( mesh, 0.0, oParams)
	if ( decimate ):
		resMesh.packOptimally(False)
		dSettings = mm.DecimateSettings()
		dSettings.maxError = 0.25 * voxelSize
		dSettings.tinyEdgeLength = mesh.computeBoundingBox().diagonal() * 1e-4
		dSettings.stabilizer = 1e-5
		dSettings.packMesh = True
		dSettings.subdivideParts = 64
		mm.decimateMesh(resMesh,dSettings)
	return resMesh

[Grantim]

Thanks for sharing! Computational complexity of this function is something about O((1/voxelSize)^3) you can try to increase voxel size for better performance. We have function mm.suggestVoxelSize( mesh, 5e6 ).

[dhanraj-khatal]

do this process require GPU based instance

[Grantim]

Not really, our python module only uses CPU computations now. We have plans to expose CUDA options too, but it is not implemented yet.