Research Interests

My primary interests lie in developing new computer rendering algorithms. Specifically, I am interested in determining novel ways handling the ever increasing amount of complexity in modern scenes. My research incorporates elements of human psychology, design, and art, as well as, the body of more traditional computer graphics work. I believe that to create truly realistic images rendering algorithms must be able to handle complexities in geometry, lighting and material orders of magnitude greater supported by current means. I believe viable solutions these problems must solve two canonical tasks. First, they must determine the visible importance of each element of a scene. This task is intrinsically related to how we seamlessly edit the enormous visual complexity of our world, and how our brains extract relevent patterns from this complexity. Second, solution must permit the computer's representation of objects, materials and lighting to evolve to reflect their relative visual importance.

Advisors

I am a student of Kavita Bala and her work is my primary focus. I also work closely with Bruce Walter, Sebastian Fernandez, and all the staff and students of Cornell's Program of Computer Graphics. Additionally, I have recently been doing some work with Steve Marschner.

Current Projects

The Lightcuts project has developed a robust rendering framework that produces full global illumination effects from up to millions of non-trivial visible lights. The methods are robust and scalable; producing high-quality anti-aliased images in times logarithmic in the number of lights.

The Hardware Environment Map project brings the benefits of image based lighting techniques to interactive hardware applications. Leveraging the recent advances in hardware shadowing algorithms, this work divides input environment maps into components that can be rendered interactively using traditional means. In practive, this division can support accurate approximations of environment shadows at interactive rates for fully dynamic scenes.

With Steve Marschner, I have been working on Wood Material Aqcuisition using his spherical gantry, a device to automates the acquisition Bidirectional Reflectance Distribution Functions (BRDFs) for materials. By fitting a new model of wood's reflectance to the data acquired for several different samples of finished hardwood, we have developed a model that accurately captures the rich, subtle appearance of many finished woods.