Processing code to evolve colors towards random RGB targets using a simple genetic algorithm

The drawing represents one generation per line, one individual per box. The bottom color is the target. It changes every n generations.

An individual is charaterized by its genotype (RGB values between 0 and 255) and expresses a phenotype (the corresponding color). The fitness function is the inverse of the Manhattan distance between the individual RGB genotype and the target RGB. Each individual has a probability to reproduce that is proportional to its fitness. A child is created from two different parents, and each of its genes is randomly selected from one parent or the other. A mutation occurs on a child genes with a probability p_mutation. If it occurs, one gene is selected and a Gaussian noise of power sigma is added to its value.

The html_version.html file allows to print the result in a html page using Processing.js.

This work is inspired by some parts of The Nature of Code, freely accessible here: http://natureofcode.com/book/

Learn more on my blog

Author: Cédric Colas

Email: cdric.colas@gmail.com