Quadric mesh simplification

A leightweight package for simplifying a mesh containing node features. The algorithm from Surface Simplification Using Quadric Error Metrics was implemented using cython.

Only python versions >= 3.6 are supported

Installation

with pip

$ pip install quad_mesh_simplify

from source if distribution is not supported

Download this repository and build the package by running:

$ pip install -r requirements.txt
$ python setup.py build_ext --inplace
$ pip install .

Usage

This package provides one simple function to reduce a given mesh. This can be done for simple meshes or meshes with vertex features.

simplify_mesh(positions, face, num_nodes, features=None, threshold=0., max_err=np.Infinity)

positions (numpy array): array of shape [num_nodes x 3] containing the x, y, z position for each node

face (numpy array): array of shape [num_faces x 3] containing the indices for each triangular face

num_nodes (int): number of nodes that the final mesh will have threshold (number, optional): threshold of vertices distance to be a valid pair

features (numpy array): features for all nodes [num_nodes x feature_length]

threshold (double): if the distance between two vertices is below this threshold, they are considered as valid pairs that can be merged.

max_err (double): no vertices are merged that have an error higher than this number. IMPORTANT: if provided it is not guaranteed that the output will have less than num_nodes vertices.

Returns: new_positions, new_face, (new_features)

Reduce a simple mesh

    from quad_mesh_simplify import simplify_mesh

    new_positions, new_face = simplify_mesh(positions, face, <final_num_nodes>)

Reduce a mesh with vertex features

    from quad_mesh_simplify import simplify_mesh

    new_positions, new_face = simplify_mesh(positions, face, <final_num_nodes>, features=features)

Reduce a mesh with a threshold for the minimal distance

    from quad_mesh_simplify import simplify_mesh

    new_positions, new_face = simplify_mesh(positions, face, <final_num_nodes>, threshold=0.5)