UNM Researchers Find Quicker, Less Expensive Way to do High Quality Rendering for Computer Generated Films
February 15, 2012
Categories: Inside UNM
Pradeep Sen, Computer and Engineering assistant professor and graduate student Soheil Darabi, have found a new way to solve an old problem in the digital film industry.
Normally it takes thousands of hours of computing time to calculate the final image frames in films that use computer graphics, such as those with special effects or completely computer-generated films.
The reason for this is that computers typically use a Monte Carlo-based algorithm (an algorithm based on random numbers) to simulate how the light bounces around the complex environments. If only a few random calculations are performed, the resulting image is very noise and does not look correct, so film makers use millions of Monte Carlo calculations to produce each final frame of film, considerably slowing down the computation process.
Sen and Darabi have found a way to perform a statistical analysis on the Monte Carlo result produced after just a few minutes in order to remove the artifacts and generate an image that is visually comparable to the one produced with millions of Monte Carlo calculations.
The advantage of their method is that it reduces the computation time by several orders of magnitude, thereby enabling the rendering of complex effects for quickly.
In this conversation with Karen Wentworth, Sen describes their work.
Sen and Darabi will present their new algorithm this summer at SIGGRAPH 2012, (Asso. for Computing Machinery Special Interest Group on Computer Graphics and Interactive Techniques) the premier venue for cutting-edge research in the field of computer graphics and digital media.
An explanation of their research and a video demonstration are available here.
Students interested in working in this area with Sen can find out more at the Advanced Graphics Lab website.
Sen and Darabi have already filed a patent disclosure through STC.UNM, a wholly owned subsidiary of the university that develops the intellectual property of Faculty, staff and students, and the technology is now available for licensing.
Media contact: Karen Wentworth (505) 277-5627; email: email@example.com