A new patent from Philips proposes a new data storage and algorithmic reconstruction for stereoscopic 3D data based on mixing a Depth Map with the video stream and allowing a compute processor to reconstruct as many stereo views as necessary.
stereoscopic data may comprise a so-called depth map that is associated with a visual image. A depth map can be regarded as a set of values, which may be in the form of a matrix. Each value relates to a particular area in the visual image and indicates a distance between a virtual observer and an object in that area. The area concerned may comprise a single pixel or a cluster of pixels, which may have rectangular shape, or any other shape for that matter. A processor is capable of generating different views, which are required for stereoscopic rendering, on the basis of the visual image and the depth map associated therewith.
At first glance this seems a bit pointless for movies, until you realize the number of Autostereoscopic displays that can represent multiple views, aside from the classic stereo pair.
