Mickey Mouse in Switzerland

In 2008 cooperation between the Walt Disney Company and the Swiss Federal Institute of Technology Zürich (ETH) was announced as part of the Disney Research group. This unusual combination of no-nonsense Swiss thinking and the nonsense-centric Disney creativity raised some eyebrows and many thought this is not going anywhere. Well, in March of 2010 "Disney Research Zürich" was officially opened with a group of 20 scientists and is poised to grow to 40 by 2011.

As nobody really knew what the goal was for this research group, the targets are now becoming clearer. In an ETH publication two main focus areas were described as modeling human faces (a holy grail for animated movies) and 3D movies. The latter one became clearer with a recent announcement. 2D editing methods for movies don’t really work well with 3D and don’t address any 3D issues. New is an algorithm that allows editors to correcting the depth of individual objects. This can be done in a post-processing step or even on the fly for live events. It may also allow converting 2D content into 3D content.

The attached 3D anaglyph pictures (from ETH Life - Ein Algorithmus für mehr 3D-Sehgenuss) show one of the key issues that exist with today’s 3D imaging.

If an object is coming out of the screen (top picture) it can create eye strain if the out-of-screen effect is too strong. It can also lead to an uncomforting viewing experience when the out-of-screen object overlaps the edge of the display (which it does). This breaks the 3D illusion by creating contradictory depth cues. In such cases, our brain has to resolve this issue through more processing. If we do this in a computer the processor, will heat up and the fan will start spinning faster to keep it cool. But in the human case, we just get a headache. This is a typical case where new technology is required to make 3D ready for the consumer not only in the movie theater but also in the home.

Disney Research is addressing this issue by making it possible to alter the depth structure at the pixel level. This allows varying the depth structure of single objects or planes of the complete scene. In the lower picture, the car is pushed back into the picture and the edge of the car is closer to screen depth so as not to have a conflict with the border edge.

Click to watch the video

More information:
Nonlinear Disparity Mapping for Stereoscopic 3D

By Norbert Hildebrand, Display Daily