Genetic Image

(11, 2012)

I wrote a program that attempts to reconstruct an image using millions of translucent triangles. It simulates evolution by spawning a generation of triangles with slight variations, and allowing only those which are the closest to the target image (i.e., the ‘fittest’) and allowing them to move on to the next generation. Over hundreds of thousands of generations, these random variations take the shape of the target image, with more accuracy the longer the program is left to run. The video above shows a target image followed by a short timelapse segment of the evolutionary process, and finally the completed image after over 12 hours of running.