Sign in to view more
Sign in to view more
Dr. John Snyder is a Principal Researcher at Microsoft Research, Redmond, WA in the 3D Graphics research group. He is leading research efforts on real-time global illumination effects such as soft shadowing and is also interested in mathematical methods and representations for rendering and processing synthetic geometry. The ambition of synthesizing all known or imaginable visual effects in real-time remains far from fully realized, in large part because such synthesis is tremendously expensive to compute. Fortunately, specialized graphics hardware is becoming more powerful and multi-core computers are also developing. Efficient graphics algorithms must be computed in parallel, respect the memory hierarchy by computing over small working sets, and minimize communication between threads and between the CPU and graphics co-processor (GPU). Dr. Snyder’s research explores new representations and algorithms that take advantage of evolving hardware capabilities to heighten the quality of real-time computer graphics. More concretely, traditional graphics renders “hard” shadows from a point light source like a candle, but neglects “soft” effects including shadows from a large area light source like a cloudy sky or a picture window. These effects add realism and provide cues that enhance our perception of shape. Capturing soft effects traditionally involves sampling rays originating from every surface point in all directions towards the light sources to determine whether intervening objects block or reflect the light. This is much too expensive for interactive applications. Along with other collaborators, Dr. Snyder’s work factors rendering into a pre-processing and a run-time phase, called precomputed radiance transfer (PRT), to make such effects practical in real time. Though the original PRT work was restricted to static geometry, he has recently been extending it to dynamic geometry, including moving characters. The work has appeared in a series of academic papers and made available in Microsoft’s DirectX API. It has been used in several popular 3D game titles including Halo 3. Another theme in Dr. Snyder’s research is the efficient storage and manipulation of surface signals, such as colors, defined over a geometric surface. Dr. Snyder introduced the “stretch” metric to parameterize surfaces and create texture maps to store these signals that are as small as possible. The idea has been developed in a series of papers and implemented in the UVAtlas tool (part of DirectX9). This tool represents the state of the art in automatic surface parameterization and has been used in many commercial 3D games. Dr. Snyder joined Microsoft Research in January 1994. Prior to that, he was a graduate student in Computer Science at the California Institute of Technology from 1984 to 1991, supervised by Al Barr and Jim Kajiya. After graduating, he remained at Caltech as a post-doctoral researcher from 1991 to 1993. At Caltech, he worked on high-quality rendering methods and techniques for modeling shape for CAD and entertainment applications. His thesis was on a new geometric modeling technique, called generative modeling, which represents shapes as multidimensional, continuous, parametric functions. A set of symbolic operators on such functions is used to build up complicated shapes by simple composition and by computational solid geometry (CSG). His thesis showed how man-made shapes can be specified using this method, and how numerous types of manipulation on them can be supported by three simple methods associated with each operator: point evaluation, symbolic differentiation, and evaluation of an inclusion function. The results were published in several papers, as well as a book “Generative Modeling for Computer Graphics and CAD”, published by Academic Press in 1992.