Sophisticated computer graphics techniques used in surface modeling, such as bidirectional texture functions (BTFs), reaction-diffusion, pattern and shape distribution, etc., can involve synthesizing and storing complex surface details. Hence, directly applying surface modeling to computer-aided design or time-critical rendering may not be adequately efficient. Using tile-based methods is a key to the problem as tiles are usually compact and reusable. Moreover, by synthesizing surface details on tiles and map them onto geometry surfaces, one can decouple surface modeling from geometries and reduce the complexity of the synthesis process. This dissertation proposes the use of tile-based methods in reaction-diffusion and Poisson disk distribution generation, and demonstrates a wide range...[ Read more ]

Sophisticated computer graphics techniques used in surface modeling, such as bidirectional texture functions (BTFs), reaction-diffusion, pattern and shape distribution, etc., can involve synthesizing and storing complex surface details. Hence, directly applying surface modeling to computer-aided design or time-critical rendering may not be adequately efficient. Using tile-based methods is a key to the problem as tiles are usually compact and reusable. Moreover, by synthesizing surface details on tiles and map them onto geometry surfaces, one can decouple surface modeling from geometries and reduce the complexity of the synthesis process. This dissertation proposes the use of tile-based methods in reaction-diffusion and Poisson disk distribution generation, and demonstrates a wide range of surface modeling applications.