Research to Improve Graphics, Digital Video, and DisplaysCategory: Science & Technology
Posted: July 30, 2013 08:45AM
Sight is one of the most valuable senses we have, as anyone who produces digital video or computer graphics will assure you. Despite those persons' best efforts though, sometimes what they create just does not look real because they do not have the tools to achieve realism. Researchers at Harvard University however have recently published three papers to help bridge gaps between reality and the media they produce.
One of these papers focuses on how computer graphics treat translucent objects. We have known for some time the math behind translucency, but understanding how it relates to what we see, so CG artists can make materials look real, would require a prohibitive amount of work. The researchers decided to take advantage of the power of modern computers to simplify the work by rendering thousands of images and comparing them, pixel by pixel, to find only those that are representative of all the images. They then showed those images to people, to learn how we interpret a material's translucency.
The two other research papers describe an adaptive display and tool to simplify a video editing technique normally limited to professionals. The adaptive display the paper describes takes advantage of a pattern etched into the display's surface, to present three-dimensional objects that change depending on the angle you view them at, like when you look through a window. The third paper discusses a new color-grading method, which could bring the editing of a frame's color palette to amateur videos. As the process currently requires frame-by-frame editing, it is largely limited to professional films with large budgets.
Source: Harvard University