Image Markov Chain Monte Carlo

imcmc

It probably makes art.

imcmc (im-sea-em-sea) is a small library for turning 2d images into probability distributions and then sampling from them to create images and gifs. Right now it is best at logos and shape based images.

Installation

This is actually pip installable from git!

pip install git+https://github.com/ColCarroll/imcmc

Quickstart

See imcmc.ipynb for a few working examples as well.

import imcmc


image = imcmc.load_image('python.png', 'L')

# This call is random -- rerun adjusting parameters until the image looks good
trace = imcmc.sample_grayscale(image, samples=1000, tune=500, nchains=6)

# Lots of plotting options!
imcmc.plot_multitrace(trace, image, marker='o', markersize=10,
                      colors=['#0000FF', '#FFFF00'], alpha=0.9);

# Save as a gif, with the same arguments as above, plus some more
imcmc.make_gif(trace, image, dpi=40, marker='o', markersize=10,
               colors=['#0000FF', '#FFFF00'], alpha=0.9, 
               filename='example.gif')