Conversions across various representations¶

Kaolin supports conversions across all popular 3D representations. Here’s a quickfire introduction.

For this tutorial, we will load the model.obj file inside of the tests directory (tests is contained in the base directory of the Kaolin repository).

>>> mesh = kal.rep.TriangleMesh.from_obj('model.obj')

As a sanity check to ensure the mesh is read in correctly, run.

>>> mesh.vertices.shape
torch.Size([482, 3])
>>> mesh.faces.shape
torch.Size([482, 3])

TriangleMesh to VoxelGrid¶

To convert a triangle mesh to a 32 x 32 x 32 voxel grid, use

>>> voxels = kal.conversions.trianglemesh_to_voxelgrid(mesh, 32)

To convert a triangle mesh to a pointcloud with 1000 points, use

>>> voxels = kal.conversions.trianglemesh_to_pointcloud(mesh, 1000)

To convert a triangle mesh to a signed distance function, use

>>> sdf = kal.conversions.trianglemesh_to_sdf(mesh)

To convert a voxel grid to a pointcloud (containing 1000 points), use

>>> sdf = kal.conversions.voxelgrid_to_pointcloud(voxels, 1000)

To convert a voxel grid to a triangle mesh, use

>>> verts, faces = kal.conversions.voxelgrid_to_trianglemesh(voxel, mode='marching_cubes')

To convert a voxel grid to a quad mesh, use

>>> verts, faces = kal.conversions.voxelgrid_to_quadmesh(voxels, thresh=.1)

To convert a voxel grid to a signed distance function (SDF), use

>>> sdf = kal.conversions.voxelgrid_to_sdf(voxels, thresh=.5)

To convert an SDF to a pointcloud, use

>>> points = kal.conversions.sdf_to_pointcloud(sdf, bbox_center=0., resolution=10, bbox_dim=1,  num_points = 10000)

To convert an SDF to a triangle mesh, use

>>> verts, faces = kal.conversions.sdf_to_trianglemesh(sdf, bbox_center=0., resolution=10, bbox_dim=1)

To convert an SDF to a voxel grid, use

>>> voxels = kal.conversions.sdf_to_voxelgrid(sdf, bbox_center=0., resolution=10, bbox_dim=1)

To convert a pointcloud to a voxel grid, use

>>> voxels = kal.conversions.pointcloud_to_voxelgrid(points, 32, 0.1)

To convert a pointcloud to a triangle mesh, use

>>> mesh_ = kal.conversions.pointcloud_to_trianglemesh(points)

To convert a pointcloud to an SDF, use

>>> sdf_ = kal.conversions.pointcloud_to_trianglemesh(points)