kaolin.conversions.meshconversions

trianglemesh_to_pointcloud(mesh: kaolin.rep.Mesh.Mesh, num_points: int)

Converts passed mesh to a pointcloud

Parameters

• mesh (kaolin.rep.Mesh) – mesh to convert

• num_points (int) – number of points in converted point cloud

Returns

converted point cloud

Return type

(torch.Tensor)

Example

>>> mesh = kal.TriangleMesh.from_obj('object.obj')
>>> points = kal.conversions.trianglemesh_to_pointcloud(mesh, 10)
>>> points
tensor([[ 0.0524,  0.0039, -0.0111],
        [-0.1995,  0.2999,  0.0408],
        [-0.1921, -0.0268,  0.1811],
        [ 0.1292,  0.0039,  0.2030],
        [-0.1859,  0.1764,  0.0168],
        [-0.1749,  0.1515, -0.0925],
        [ 0.1990,  0.0039, -0.0083],
        [ 0.2173, -0.1285, -0.2248],
        [-0.1916, -0.2143,  0.2064],
        [-0.1935,  0.2401,  0.1003]])
>>> points.shape
torch.Size([10, 3])

trianglemesh_to_voxelgrid(mesh: kaolin.rep.Mesh.Mesh, resolution: int, normalize: bool = True, vertex_offset: float = 0.0)

Converts mesh to a voxel model of a given resolution

Parameters

• mesh (kaolin.rep.Mesh) – mesh to convert

• resolution (int) – desired dresolution of generated voxel array

• normalize (bool) – Determines whether to normalize vertices to a unit cube centered at the origin.

• vertex_offset (float) – Offset applied to all vertices after normalizing.

Returns

voxel array of desired resolution

Return type

voxels (torch.Tensor)

Example

>>> mesh = kal.TriangleMesh.from_obj('model.obj')
>>> voxel = kal.conversions.trianglemesh_to_voxelgrid(mesh, 32)
>>> voxel.shape

trianglemesh_to_sdf(mesh: kaolin.rep.Mesh.Mesh, num_points: int = 10000)

Converts mesh to a SDF function

Parameters

• mesh (kaolin.rep.Mesh) – mesh to convert.

• num_points (int) – number of points to sample on surface of the mesh.

Returns

a signed distance function

Return type

sdf

Example

>>> mesh = kal.TriangleMesh.from_obj('object.obj')
>>> sdf = kal.conversions.trianglemesh_to_sdf(mesh)
>>> points = torch.rand(100,3)
>>> distances = sdf(points)