k402xxxcenxxx / PyVoxSurf

A Python C++ extension based on VoxSurf for voxelization of 3D meshes.

PyVoxSurf

A Python C++ extension based on VoxSurf for voxelization of 3D meshes.

NOTE: PyPi distribution is only compatible with Microsoft Windows

Principle

1. Rasterize triangles using three 2D axis aligned grids, using integer arithmetic (fixed floating point) for robust triangle interior checks
2. [Optional] Fill interior of voxelized surface with either of two schemes: Inside - fastest method evaluates whether a voxel is inside from only one direction or Robust - evaluates whether a voxel is inside from all three directions and a voting determines final status

Usage

pyvoxsurf.voxelize_stl

Argument	Type	Default	Description
filename	string		Filename of .stl file
resolution	integer		Number of voxel slices in the z-axis
bounds	[2x3] array		[Optional] Min and max bounds in (x, y, z) coordinates of desired voxel volume
voxel_fill	string	"None"	[Optional] "None", "Inside", or "Robust" type of filling

import pyvoxsurf
from mayavi import mlab

volume1 = pyvoxsurf.voxelize_stl("model.stl",200,[],"Robust")
print(volume1.shape)

# Visualize voxelized model
from tvtk.util.ctf import PiecewiseFunction
mlab.figure(size=(800,800))
vol = mlab.pipeline.volume(mlab.pipeline.scalar_field(volume1))
mlab.title('Voxelized model',height=0.9,size=0.5)
mlab.orientation_axes()
otf = PiecewiseFunction()
otf.add_point(0,0)
otf.add_point(0.001, 1)
otf.add_point(1,1)
vol._otf = otf
vol._volume_property.set_scalar_opacity(otf)
mlab.show()

pyvoxsurf.voxelize

Argument	Type	Default	Description
vertices	[nx3] array		Vertex positions in (x, y, z) coordinates
triangle_indices	[nx3] array		Indices of connected vertices forming triangles of mesh
bounds	[2x3] array		Min and max bounds in (x, y, z) coordinates of desired voxel volume
resolution	integer		Number of voxel slices in the z-axis
voxel_fill	string	"None"	[Optional] "None", "Inside", or "Robust" type of filling

import pyvoxsurf
import trimesh
import numpy as np
from mayavi import mlab

mesh = trimesh.load("model.stl") # Load stl file

# Find the max and min coordinates of the mesh to form a bounding box
mesh_min_corner = [np.min(mesh.vertices[:,0]), np.min(mesh.vertices[:,1]), np.min(mesh.vertices[:,2])]
mesh_max_corner = [np.max(mesh.vertices[:,0]), np.max(mesh.vertices[:,1]), np.max(mesh.vertices[:,2])]
bounds = np.stack((mesh_min_corner,mesh_max_corner))

volume2 = pyvoxsurf.voxelize(mesh.vertices,mesh.faces,bounds,100,"Inside")
print(volume2.shape)

# Visualize voxelized model
from tvtk.util.ctf import PiecewiseFunction
mlab.figure(size=(800,800))
vol = mlab.pipeline.volume(mlab.pipeline.scalar_field(volume2))
mlab.title('Voxelized model',height=0.9,size=0.5)
mlab.orientation_axes()
otf = PiecewiseFunction()
otf.add_point(0,0)
otf.add_point(0.001, 1)
otf.add_point(1,1)
vol._otf = otf
vol._volume_property.set_scalar_opacity(otf)
mlab.show()

Credits

• VoxSurf by sylefeb
• VoxSurf Pybind11 bindings by mjgalindo
• PyVoxSurf packaging and documentation by jttoombs
• STL model of 3D knot by chylld