Weighted Voronoi Stippling Mesher

Information

• Author: Evan John Ricketts
• Afffiliation: Cardiff University
• Date: April 27, 2024

Description

Mesh generator based on weighted voronoi stippling, where an image is taken as input, and its luminosity dictates mesh density.

This can then be used in a number of ways, such as a finite element solver. Below is the solution of Poisson's equation with two souce terms.

Sources of inspiration

• [1] Journal article: Paper link - Secord, A. (2002). Weighted Voronoi stippling. In Proceedings of the 2nd international symposium on Non-photorealistic animation and rendering. NPAR02: Non-Photorealistic Animation and Rendering. ACM. https://doi.org/10.1145/508530.508537

• [2] Github repo: Weighted voronoi stipplng - Jannis Vamvas

• [3] Github repo: Particle Repulsion Mesher - Vlad Yastrebov

• [4] YouTube video: Coding Challenge 181: Weighted Voronoi Stippling

Requirements

• Python 3.7 or higher
• Numpy
• Numba
• Matplotlib
• PIL
• Scipy
• Tqdm

Usage

The code -with default arguments- can be run from the command line as follows:

python main.py image_name.png

where it is assumed the images is in the images directory.

The following parameters can be parsed

    npoin = 30000       # Number of particles
    niter = 25          # Number of iterations
    save = False        # Save array of final points
    animate = False     # Animate the meshing process
    plot = False        # Plot final mesh
    dpi = 400           # Resolution of plots

Examples

Strawberry

Command: python main.py sbry.jpg --npoin 20000 --niter 25 --plot

Random Field 3

Command: python main.py RF3.png --npoin 20000 --niter 25 --plot

Skull

Command: python main.py skull.png --npoin 20000 --niter 25 --plot

Note

The code is provided as is and the author is not responsible for any damage caused by the code. GPT4 was utilised in the construction of the code.