Genetic algorithms in animation of volumetric clouds

Abstract

Realistic modeling and animating clouds are ones of the hardest tasks in computer animation. There are different techniques used in computer graphics for representing clouds. Our work is based on volume-rendered implicit junction, for volumetric cloud modeling. Basically, implicit primitives are controlled by a modified particle system. We propose genetic algorithms for initialization of the particle system. They provide an alternative to traditional optimization techniques and are also useful in the key-frame animation. Additionally, we introduce some new ideas in volumetric clouds animation: advanced techniques for controlling a large collection of individual elements - flocks and autonomous behavior. Some properties of these techniques are also applicable in the cloud animation.