Synthetic images have reached considerable sophistication, to the point where we can render images indistinguishable from reality. Today the major limiting factor for a ubiquitous use of computer-generated images is the human labor and expertise required to create the shape, materials and lights of synthetic environments. This project is a combined research and education effort that brings us closer to making the creation the synthetic imagery accessible to all. The research component of this project simplifies the design of objects' appearance, which comes from the interaction of materials and lights, to complement recent advances in shape modeling and animation. The goal is to allow users, including novices, to design the appearance of complex scenes in just minutes. On the education side, the project explores the interaction between the conceptual, technical, and aesthetic principles of image synthesis through curriculum development and out-of-classroom experiences.
More specifically, the project investigates interfaces that allow designers to intuitively and effectively specify design goals on objects? appearance, algorithms that derive lights and materials parameters robustly from such goals, and representations of appearance that are effective to manipulate. These investigations allow designers to manipulate complex lighting and materials with intuitive user-interface metaphors and to transfer appearance from example images. Qualitative and quantitative user studies guide our investigation and serve as rigorous validation of our results. We focus on novice users, but expect our work to benefit experts as well. The resulting methods allow intuitive and fast modeling, while remaining consistent across all aspects of appearance design, from simple lighting and materials to complex environmental illumination and textured surfaces, in realistic and non-photorealistic renderings of static and dynamic scenes.
