This couldn't have come at a better time (for me). I'm playing with a genetic programming library to recreate images from some basic rules and I've been really unhappy with the fitness equality functions I've been able to find. This looks like it may have been made for the job.