Modeling natural phenomena with implicit surfaces

In this dissertation a computer graphics modeling and animation environment is proposed, which is well suited for modeling complex natural phenomena. The system is based on a data structure called the BlobTree, and employs implicit surfaces as the modeling paradigm. Implicit surfaces offer an inherent advantage over traditional surface modeling methods for many natural phenomena due to the wide array of intuitive blending techniques available. To alleviate the slow visualization times imposed by implicit surfaces, new methods for accelerated visualization of BlobTree models are presented. Model specification employs a procedural interface allowing specification of animation parameters and the inclusion of physically based simulations. An interactive viewing program is then used to preview the models using approximations with varying degrees of accuracy and speed, and to ray trace models for photo-realistic results. The proposed system supports the construction of implicit surface models containing several orders of magnitude more primitives than has previously been possible. The use of this system is demonstrated by building models of several complex natural phenomena including a Murex cabritii shell, an animation of 27 years of growth for a Populus deltoides tree, and a physically based simulation of ice and speleothem formation. Each of these examples represent improved visual realism in the resulting models compared to previous techniques.