SCS Undergraduate Thesis Topics
|Andrew Cove||Doug James||Dynamic Texturing of Botanical Environments|
Data-driven animation techniques enable real-time synthesis of physically based motions without the cost of run-time simulation. The overhead costs of precomputing motion and lighting for a variety of objects, behaviors, and environmental conditions are large, but tolerable given the performance benefits at run-time. Since a primary goal of physically based modeling is to synthesize real world environments, a natural extension of precomputing and synthesizing complex behaviors for simulated objects is to place the objects in scenes representative of their natural environments. However, the cost of assembling and precomputing physical states for an environment scales as the environment grows in size and complexity, and the reuseability of the data diminishes. We provide techniques for synthesizing expansive, interactive environments that maintain the low run-time costs of data-driven animation without incurring the overhead costs of building and simulating large scenes. We utilize a small library of data-driven irises and an extension of Wang Tiles to generate a large field of irises that can be animated in real-time under varying wind and lighting conditions.