A new technique for faster graphics

Aug 13, 2007 14:54 GMT  ·  By

Generating complex and realistic looking images, rendering as it is called, currently involves a lot of memory space and processor time from both the central and the graphics processors. The situation is even more complex and more hardware intensive when we are talking about good looking images that involve one kind or another of smoke, fog or any other mediums that absorb or reflect light.

According to the news site vnunet.com, a team of researchers from the United States developed a method of rendering complex and real looking images like smoke filled rooms or landscapes that involves just a fraction of the computing power needed when using traditional approaches. "This is a huge computational saving. It lets you render explosions, smoke, and the architectural lighting design in the presence of these kinds of visual effects much faster," said Wojciech Jarosz, the UC San Diego computer science PhD candidate who led the study. "Being able to accurately and efficiently simulate these kinds of scenes is very useful".

As with any other complex images that are rendered, the smoke and fog effects are largely based on the fact that different surfaces and mediums will reflect, absorb and distort the light in a different and particular way. This new approach takes the things a little bit further and when a scene is encountered that involves smoke, clouds, fog or any other participating media, the lightning scheme is calculated accurately for a small number of locations within the generated image and then based on the known light values, interpolations are made for all nearby points. This method is not really new as it can be considered an extension of the 'irradiance caching' filtering method.

The good part of it is that using such an approach the number of computations that are needed alongside a certain line of sight is drastically reduced. Of course, inside a 3D scene, moving from one point to another or modifying the viewing angle will prompt a new chain of calculations, but even so the total hardware "pressure" is much smaller.