The use of an aesthetic measure for the evolution of fractal landscapes

Abstract

This paper explores the use of an aesthetic measure to aid the generation of fractal landscapes. Virtual landscapes are important for applications ranging from games to simulation. This paper extends work done on the auto generation of virtual landscapes for climate change visualisation, by adding an aesthetic measure based fitness function to the evolutionary algorithm, thus reducing the reliance of the method on user based evaluation. A genetic algorithm that uses an aesthetic measure of fitness based on information theory is defined. This GA is used to explore a multi-dimensional parameter space that defines how 3D virtual landscapes are created. The utility of this fitness measure is assessed by evaluating the solutions generated by the system with real users. Results indicate that genetic algorithms that use information theory based fitness measures do indeed generate virtual artefacts that match user preferences. Moreover the images generated are visually appealing enough to be curated for public exhibition along side human artists in art galleries by professional art critics.