ÌÇÐÄTV

Abstract:

One of the main goals of Computer Graphics is rendering realistic images in real time. Most of my current research effort is devoted towards achieving this goal. I will present a part of my recent work on image space approaches to realistic rendering of Caustics and Subsurface scattering in real time. In realistic rendering, we come across many problems that are expensive to solve in 3D object space. We show that image-space approaches can provide effective solutions to these problems. Unlike object space algorithms that operate on 3D primitives such as points, edges, and polygons, image-space algorithms operate on 2D snapshot images of the 3D geometric data. Operating in image-space effectively decouples the geometric complexity of the 3D data from the run-time of the rendering algorithm. Other important advantages of image-space algorithms include ease of implementation on modern graphics hardware, and fast computation of approximate solutions to certain lighting calculations. We have applied this image-space approach for rendering caustics, a subtle effect resulting from reflection and refraction of light on complex objects, and rendering translucency that results from subsurface scattering of light on translucent objects.