However, polygon throughput may no longer be the only measure of performance. The simple "toy" environments of the eighties, which had many pixels per polygon, allowed algorithms to rely on polygon coherence to simplify graphics processing. Now we render models with hundreds of thousands (or even millions) of objects, having many polygons per pixel. While hardware optimized for pumping out triangles is impressive, new approaches are needed to render complex dynamic environments in real time.

Cornell's Program of Computer Graphics is conducting a leading research effort in realistic image synthesis. We are advancing our algorithmic techniques to achieve unprecedented speed in calculating global illumination in synthetic scenes. At the same time, we are exploring new techniques including digital photographic image processing, image acquisition, and image based rendering.

Realism is still key if we want our simulations to be predictive and to have value beyond illustration and entertainment. We have unique resources to pursue the physical basis of light reflection as an integral foundation for our algorithms. We are also pioneering research in visual perception, recognizing the greatest advances in speed will come from limiting our computations to those producing differences perceptible to the human eye.

Architectural modeling, particularly at the early stages of design, represents one of the most difficult challenges for the interactive graphics software industry. The Program of Computer Graphics is a leading research lab in interactive modeling and is also teaching an innovative architectural design studio.

Cornell is known for graduate education that crosses traditional boundaries. Our lab is a multidisciplinary center with ties to Architecture, Art, Mechanical Engineering and Psychology as well as Computer Science. Our graduate students participate on research teams with faculty and students from other disciplines, stretching their own ideas to address new challenges.