hub/venv/lib/python3.7/site-packages/trimesh/permutate.py

153 lines
4.5 KiB
Python

"""
permutate.py
-------------
Randomly deform meshes in different ways.
"""
import numpy as np
from . import transformations
from . import triangles as triangles_module
from . import util
def transform(mesh, translation_scale=1000.0):
"""
Return a permutated variant of a mesh by randomly reording faces
and rotatating + translating a mesh by a random matrix.
Parameters
----------
mesh : trimesh.Trimesh
Mesh, will not be altered by this function
Returns
----------
permutated : trimesh.Trimesh
Mesh with same faces as input mesh but reordered
and rigidly transformed in space.
"""
# rotate and translate randomly
matrix = transformations.random_rotation_matrix()
matrix[0:3, 3] = (np.random.random(3) - 0.5) * translation_scale
# randomly re-order triangles
triangles = np.random.permutation(mesh.triangles).reshape((-1, 3))
# apply rigid transform
triangles = transformations.transform_points(triangles, matrix)
# extract the class from the input object
mesh_type = util.type_named(mesh, 'Trimesh')
# generate a new mesh from the permutated data
permutated = mesh_type(
**triangles_module.to_kwargs(triangles.reshape((-1, 3, 3))))
return permutated
def noise(mesh, magnitude=None):
"""
Add gaussian noise to every vertex of a mesh, making
no effort to maintain topology or sanity.
Parameters
----------
mesh : trimesh.Trimesh
Input geometry, will not be altered
magnitude : float
What is the maximum distance per axis we can displace a vertex.
If None, value defaults to (mesh.scale / 100.0)
Returns
----------
permutated : trimesh.Trimesh
Input mesh with noise applied
"""
if magnitude is None:
magnitude = mesh.scale / 100.0
random = (np.random.random(mesh.vertices.shape) - .5) * magnitude
vertices_noise = mesh.vertices.copy() + random
# make sure we've re- ordered faces randomly
triangles = np.random.permutation(vertices_noise[mesh.faces])
mesh_type = util.type_named(mesh, 'Trimesh')
permutated = mesh_type(**triangles_module.to_kwargs(triangles))
return permutated
def tessellation(mesh):
"""
Subdivide each face of a mesh into three faces with the new vertex
randomly placed inside the old face.
This produces a mesh with exactly the same surface area and volume
but with different tessellation.
Parameters
------------
mesh : trimesh.Trimesh
Input geometry
Returns
----------
permutated : trimesh.Trimesh
Mesh with remeshed facets
"""
# create random barycentric coordinates for each face
# pad all coordinates by a small amount to bias new vertex towards center
barycentric = np.random.random(mesh.faces.shape) + .05
barycentric /= barycentric.sum(axis=1).reshape((-1, 1))
# create one new vertex somewhere in a face
vertex_face = (barycentric.reshape((-1, 3, 1))
* mesh.triangles).sum(axis=1)
vertex_face_id = np.arange(len(vertex_face)) + len(mesh.vertices)
# new vertices are the old vertices stacked on the vertices in the faces
vertices = np.vstack((mesh.vertices, vertex_face))
# there are three new faces per old face, and we maintain correct winding
faces = np.vstack((np.column_stack((mesh.faces[:, [0, 1]], vertex_face_id)),
np.column_stack(
(mesh.faces[:, [1, 2]], vertex_face_id)),
np.column_stack((mesh.faces[:, [2, 0]], vertex_face_id))))
# make sure the order of the faces is permutated
faces = np.random.permutation(faces)
mesh_type = util.type_named(mesh, 'Trimesh')
permutated = mesh_type(vertices=vertices,
faces=faces)
return permutated
class Permutator:
def __init__(self, mesh):
"""
A convenience object to get permutated versions of a mesh.
"""
self._mesh = mesh
def transform(self, translation_scale=1000):
return transform(
self._mesh, translation_scale=translation_scale)
def noise(self, magnitude=None):
return noise(self._mesh, magnitude)
def tessellation(self):
return tessellation(self._mesh)
try:
# copy the function docstrings to the helper object
Permutator.noise.__doc__ = noise.__doc__
Permutator.transform.__doc__ = transform.__doc__
Permutator.tessellation.__doc__ = tessellation.__doc__
except AttributeError:
# no docstrings in Python2
pass