My main research interest is computer graphics. I have done work in physically based image synthesis. This involves several different subjects.

Global illumination:
The propagation of light or energy throughout the environment.
Reflectance models:
The way a material reflects incident light back into the environment.
The projection of an environment into an image.
Modeling natural phenomena:
Understanding of the physical principles that govern the nature to model clouds, water, snow, atmosphere...

Current Active Projects

Outdoor rendering:
Rendering outdoor scenes is very different from indoor scenes. Usually only the sun and sky provide illumination, and the atmosphere can have a significant effect on the transport.
Approximate methods for light transport and illumination volumetric materials:
Appearance of many materials and objects cannot be represented by a simple bidirectional reflectance distribution function (BRDF). Many of the subtle effects are due to subsurface and multiple scattering. While light transport equation describes the appearance, brute force computation and simulation of appearance is prohibitively slow and expensive. Research has been done on analytic approximations that capture appearance of many natural materials: water, clouds, skin, etc.
Monte Carlo methods for Global Illumination:
Simulating realistic lighting and rendering complex scenes are usually considered separate problems with incompatible solutions. Monte Carlo methods are simple yet powerful techniques for solving global illumination equation. Metropolis sampling method first introduced in computational physics is of great simplicity and power. It can be used to sample any density function regardless of analytic complexity in any number of dimensions.
Particle-based methods for simulating natural phenomena:
Many natural phenomena can be described in terms of the Navier-Stokes equation that accurately describe fluid motion. Solving the Navier-Stokes equations numerically require resorting to different discretization methods. Gridless-particle based methods are used to define differential operators require to solve many partial-differential equations.
Digital Cartography:
How to utilize the vast amount of data avilable to display information in the most meaningful and informative way? DEM data and orthophoto are used to extract additional information from the raw data to create panorama maps.

Past Research

Liquid Crystal Simulations and Visualization: