Interactive rendering is crucial to virtual design and prototyping of large engineering models. This research will produce techniques to significantly speed up the rendering of surface models on current graphics systems. The emphasis will be on novel algorithmic improvements in processing and rendering of large models. The effectiveness of using analytic surfaces and other higher order representations of large models will also be investigated. Such representations contain more information than polygonal representations, but take less space to store. Recent improvements in efficient tessellation and rendering of parametric surfaces have opened up new possibilities for speeding up the rendering of large surface models. This research will use and improve these results. In particular, it will result in the following: - Higher-order and hierarchical representations tessellated on- line into a small number of triangles, based on the user's position - Efficient visibility processing of surfaces - Interactive rendering of dynamically changing surfaces - Improved sampling of surfaces for the generation of textured polygonal approximations - Configurable system that adjusts rendering speed and quality without requiring user input - Algorithms that scale well with both available computing resources and model size - A unified framework that allows various techniques to inter-operate and results in combined improvement in the state of the art The project also includes design and installation of new computer graphics courses and laboratories in the computer science department at the Johns Hopkins University. Significant interactions between the classrooms and laboratories are planned. The usenet and the world wide web will be widely used to augment the classroom teaching. In addition, seminars and informal meetings will be organized to discuss research in computer graphics.
