This project was built with Python. The libraries I used include:

• SciPy, to generate the Delaunay Mesh
• NumPy, to process image data and for NumPy arrays
• Pandas, to use data frames
• SciKit Image, to determine areas of high entropy on the image
• Bokeh, for everything user interface/display related

To try this project out for yourself, download the source and then run:

bokeh serve bokeh-app/

Then go to the local instance of the Bokeh Server and play around for yourself.

Warning: If you try to generate over 300 "smart points" the app will take a long time to complete all of the processing.

Make sure you are using Python 2.

An image of a face with 75 random points

An image of a face with 1009 random points

An image of a dock and lake with 3812 random points

A close up of the generated image

An image of a face with 147 smart points

Making this in Python was rather simple because of the libraries available and the description provided by this blog post

However, this implementation is ineffecient and hard to deploy (despite using a bokeh server). I'll try to continue with this project using JavaScript and writing the smart point generation more effeciently.