Procedural Mesh Generation¶

Box generation¶

pymesh.generate_box_mesh(box_min, box_max, num_samples=1, keep_symmetry=False, subdiv_order=0, using_simplex=True)¶

Generate axis-aligned box mesh.

Each box is made of a number of cells (a square in 2D and cube in 3D), and each cell is made of triangles (2D) or tetrahedra (3D).

Parameters:

box_min (numpy.ndarray) – min corner of the box.
box_max (numpy.ndarray) – max corner of the box.
num_samples (int) – (optional) Number of segments on each edge of the box. Default is 1.
keep_symmetry (bool) – (optional) If true, ensure mesh connectivity respect all reflective symmetries of the box. Default is true.
subdiv_order (int) – (optional) The subdivision order. Default is 0.
using_simplex (bool) – If true, build box using simplex elements (i.e. triangle or tets), otherwise, use quad or hex element.

Returns:

A box Mesh. The following attributes are defined.

cell_index: An numpy.ndarray of size \(N_e\) that maps each element to the index of the cell it belongs to. \(N_e\) is the number of elements.

Sphere generation¶

pymesh.generate_icosphere(radius, center, refinement_order=0)¶

Generate icosphere (subdivision surface of a regular icosahedron).

Parameters:	radius (`float`) – Radius of icosphere. center (`numpy.ndarray`) – Sphere center. refinement_order (`int`) – (optional) Number of refinement.
Returns:	The (possibly refined) icosphere `Mesh`.

Cylinder generation¶

pymesh.generate_cylinder(p0, p1, r0, r1, num_segments=16)¶

Generate cylinder (or conical frustum to be precise).

Parameters:	p0 (`np.ndarray`) – Bottom center. p1 (`np.ndarray`) – Top center. r0 (`float`) – Bottom radius. r1 (`float`) – Top radius. num_segments (`int`) – Number of segments to a discrete circle consists.
Returns:	The cylinder `Mesh`.

Tube generation¶

pymesh.generate_tube(p0, p1, r0_out, r1_out, r0_in, r1_in, num_segments=16, with_quad=False)¶

Generate generalized tube (i.e. cylinder with an axial hole).

Parameters:	p0 (`np.ndarray`) – Bottom center. p1 (`np.ndarray`) – Top center. r0_out (`float`) – Bottom outer radius. r1_out (`float`) – Top outer radius. r0_in (`float`) – Bottom inner radius. r1_in (`float`) – Top inner radius. num_segments (`int`) – Number of segments to a discrete circle consists. with_quad (`bool`) – Output a quad mesh instead.
Returns:	A generalized tube `Mesh`.

Other Platonic solids¶

pymesh.generate_regular_tetrahedron(edge_length=1.0, center=[0.0, 0.0, 0.0])¶

Generate a regular tetrahedron with the specified edge length and center.

Parameters:	edge_length (`float`) – edge length of the regular tet. center (`numpy.ndarray`) – center of the tet.
Returns:	A `Mesh` object containing a regular tetrahedron.

pymesh.generate_dodecahedron(radius, center)¶

Generate a regular dodecahedron.

Parameters:	radius (`float`) – Radius of the shape. center (`numpy.ndarray`) – shape center.
Returns:	The dodecahedron `Mesh` object.

Wire mesh generation¶

class pymesh.wires.WireNetwork¶

Data structure for wire network.

A wire network consists of a list of vertices and a list of edges. Thus, it is very similar to a graph except all vertices have their positions specified. Optionally, each vertex and edge could be associated with one or more attributes.

dim¶: int – The dimension of the embedding space of this wire network. Must be 2 or 3.

vertices¶: numpy.ndarray – A V by dim vertex matrix. One vertex per row.

edges¶: numpy.ndarray – A E by 2 edge matrix. One edge per row.

num_vertices¶: int – The number of vertices (i.e. V).

num_edges¶: int – The number of edges (i.e. E).

bbox¶: numpy.ndarray – A 2 by dim matrix with first and second rows being the minimum and maximum corners of the bounding box respectively.

bbox_center¶: numpy.ndarray – Center of the bbox.

centroid¶: numpy.ndarray – Average of all vertices.

wire_lengths¶: numpy.ndarray – An array of lengths of all edges.

total_wire_length¶: float – Sum of wire_lengths.

attribute_names¶: list of str – The names of all defined attributes.

classmethod create_empty()¶: Handy factory method to create an empty wire network.

classmethod create_from_file(wire_file)¶: Handy factory method to load data from a file.

classmethod create_from_data(vertices, edges)¶

Handy factory method to create wire network from vertices and edges.

Example

>>> vertices = np.array([
...     [0.0, 0.0, 0.0],
...     [1.0, 0.0, 0.0],
...     [1.0, 1.0, 0.0],
...     ]);
>>> edges = np.array([
...     [0, 1],
...     [0, 2],
...     [1, 2],
...     ]);
>>> wires = WireNetwork.create_from_data(vertices, edges);

load(vertices, edges)¶

Load vertices and edges from data.

Parameters:	vertices (`numpy.ndarray`) – `num_vertices` by `dim` array of vertex coordinates. faces (`numpy.ndarray`) – `num_edges` by 2 array of vertex indices.

load_from_file(wire_file)¶

Load vertices and edges from a file.

Parameters:	wire_file (`str`) – Input wire file name.

The file should have the following format:

# This is a comment
v x y z
v x y z
...
l i j # where i and j are vertex indices (starting from 1)
l i j
...

load_from_raw(raw_wires)¶: Load vertex and edges from raw C++ wire data structure.

write_to_file(filename)¶: Save the current wire network into a file.

scale(factors)¶

Scale the wire network by factors

Parameters:	factors – scaling factors. Scale uniformly if `factors` is a scalar. If `factors` is an array, scale each dimension separately (dimension `i` is scaled by `factors[i]`).

offset(offset_vector)¶

Offset vertices by per-vertex offset_vector.

Parameters:	offset_vector (`numpy.ndarray`) – A \(N imes dim\) matrix representing per-vertex offset vectors.

center_at_origin()¶: Translate the wire networks to have its center at the origin.

trim()¶: Remove all hanging edges. e.g. edge with at least one vertex of valance <= 1

filter_vertices(to_keep)¶: Remove all vertices other than the ones marked with to_keep. Edges are updated accordingly.

filter_edges(to_keep)¶: Remove all edges unless marked with to keep. Vertices are left unchanged.

compute_symmetry_orbits()¶

Compute the following symmetry orbits:

vertex_symmetry_orbit: all vertices belonging to the same orbit can be mapped to each other by reflection with respect to planes orthogonal to axis.
vertex_cubic_symmetry_orbit: all vertices belonging to the same orbit can be mapped to each other by reflection with respect to any symmetry planes of a perfect cube.
edge_symmetry_orbit: all edges belonging to the same orbit can be mapped to each other by reflection with respect to planes orthogonal to axis.
edge_cubic_symmetry_orbit: all edges belonging to the same orbit can be mapped to each other by reflection with respect to any symmetry planes of a perfect cube.

All orbits are stored as attributes.

get_attribute_names()¶: Get the names of all defined attributes.

has_attribute(name)¶: Check if an attribute exists.

add_attribute(name, value=None, vertex_wise=True)¶

Add a new attirbute.

Parameters:	name (`str`) – Attribute name. value (`numpy.ndarray`) – N by d matrix of attribute values (one row per vertex or per edge). Default is None to represent uninitialized attribute value. vertex_wise (`bool`) – Whether this attribute is assigned to vertices or edges.

get_attribute(name)¶: Get the value of an attribute.

is_vertex_attribute(name)¶: Returns true if name is a per-vertex attribute.

set_attribute(name, value)¶: Set the value of the attribute name to be value.

get_vertex_neighbors(i)¶: Returns a list of vertex indices which are connected to vertex i via an edge.

class pymesh.wires.Parameters(wire_network, default_thickness=0.5)¶: This class is a thin wrapper around PyMesh.ParameterManager class.

class pymesh.wires.Tiler(base_pattern=None)¶

class pymesh.wires.Inflator(wire_network)¶

set_profile(N)¶: Set the cross section shape of each wire to N-gon.

set_refinement(order=1, method='loop')¶

Refine the output mesh using subdivision.

Parameters:	order – how many times to subdivide. mehtod – which subdivision scheme to use. Options are `loop` and `simple`.