This is an attempt at generating an output image from a source picture by mutating a group of semi-transparent polygons.
I wanted to do something similar to this: http://rogeralsing.com/2008/12/07/genetic-programming-evolution-of-mona-lisa/
The code is still very thrown together but it produces some acceptable results. Generating the image above took a couple of hours though so there's room for improvements.
I'd be happy to hear about ideas on how to improve the algorithm/code.
-
Create virtualenv
virtualenv . -
Install requirements
bin/pip install -r requirements
-
(optional) PIL might have to be installed like this
bin/pip install PIL --allow-external PIL --allow-unverified PIL --upgrade
-
Running the script
bin/python src/generate.py <path_to_img>
Use a small source image (e.g. with a maximum size of 512x512 pixels). The generated image files will be created in your system's temp directory.