Nowadays, it is already very difficult to distinguish between real images and computer-generated graphics. However, in the near future it will be even more difficult to do so, thanks to a new algorithm developed by researchers, which allows to obtain maximum realism of light reflected from the complex surfaces of various materials, such as water, leather, glass and metal. In this case, the new algorithm performs the task of rendering approximately 100 times faster than any of the existing similar systems.
Most computer graphics production methods resort to artificial smoothing of complex surfaces in order to reduce the number of calculations and speed up processing. Such approaches were widely used in the 1980s, when the computing machinery had low performance, and the lack of detail surfaces allowed a person to immediately distinguish the computer graphics from the real images.
The new algorithm, developed by Professor Ravi Ramamoorthi’s team at the University of California, San Diego, with the help of specialists from Autodesk, gives more realistic results because it breaks down each pixel of a complex surface into a large number of so-called “microaspects. Each of these micro-aspects acts like a kind of smooth tiny mirror, reflecting light in a specific direction. And, the cooperation of tens or thousands of these micro-mirrors makes it possible to obtain a realistic image of even the most complex surface.
The technology of splitting into microaspects was already used in some rendering systems, but their processing with sufficient accuracy required “grinding” a huge number of numbers. The new system developed by the scientists reduces by a factor of 100 the number of calculations required and reduces the computer hardware requirements by 40 percent than the computer requirements needed to calculate technologies based on simplifying and smoothing surfaces.
To calculate each micro-aspect, the system needs to calculate the value of the so-called “normal vector,” which is perpendicular to the mirror surface. And, knowing in advance the value of this vector, you can accurately and quickly calculate the direction in which the light from each source will be reflected by a given micro-aspect. A virtual camera located at a certain point of the scene will perceive only those reflected rays of light that fall into its lens.
Traditionally, rendering systems calculate light reflections sequentially, from each pixel individually, which requires large computational resources. However, the Californian researchers went a different way, they grouped the micro-aspects into sections and calculated the approximate amount of light reflected by each section in a particular direction. As a result, the new algorithm was 100 times faster than before.
The next step in the development of the new rendering algorithm will be a new technology for representing complex surfaces, which will provide ultra-high resolution without requiring a drastic increase in the number of calculations and the amount of computer memory used.