Abstract

A number of evolutionary systems exist that allow the generation of esthetically pleasing images and animation. Most of these systems rely on "user-centric" fitness functions based on interactive selection. Although interactive selection is a useful tool that allows for subjective fitness evaluation, it is also restrictive as it works with a small population of individuals, and leads to a process where the user has no detailed quantitative control. We investigate a technique that enables generation of pleasing visual results that don't rely on interactive selection. The technique follows a simple principle, visual representation.