The live 3D TV system consists of (left) an array of 64 cameras that captures multi-view videos of a live scene and (right) an integral photography display with 60 viewing directions that reproduces an autostereoscopic 3D video of the scene.
(PhysOrg.com) -- Three-dimensional TV is now closer than ever to becoming a reality for consumers, and the latest research is investigating the full extent of 3D TV’s possibilities. In a recent study, researchers at the University of Tokyo and Hitachi, Ltd., have presented a 3D TV system that captures a live scene in real time and reproduces it on an autostereoscopic display. The system also offers interactive control, allowing viewers to adjust viewing parameters such as cropping a scene and reproducing an appropriate amount of depth.
“The greatest advantage of our system is to provide interactive control of the viewing parameters,” lead author of the study Yuichi Taguchi, a Ph.D. student at the University of Tokyo, told PhysOrg.com. “The interactive control is essential for reproducing a dynamic 3D scene with desirable conditions, which depend on the contents of the scene, the viewer's preference, and the display specifications.”
The 3D TV system, called TransCAIP, captures a live scene using an array of 64 video cameras that are all connected via Ethernet cables to one PC, which converts images from all the video cameras into images for the display. Each video camera contains a built-in HTTP server, which sends motion JPEG sequences to the PC.
The PC then converts the 64 input views captured by the cameras to an “integral photography image” made of 60 views, which correspond to the viewing directions of the display. Using an image-based rendering technique, the PC converts the images in real-time, and then arranges the pixels to produce the integral photography image. The entire process, called light field conversion, is implemented on the single PC in real-time, requiring a few hundred milliseconds per frame.
Like all autostereoscopic displays, the new 3D TV system doesn’t require viewers to wear special glasses. Instead, the display reproduces various viewpoint images, allowing viewers to see a different image in each eye or by moving their head (the parallax effect). Although the basic principles of autostereoscopic 3D displays were developed more than a century ago, only with recent technological advances has it been possible to actually build autostereoscopic displays since they require such a large number of light rays (the resolution of a view times the number of viewpoints).
“The greatest advantage of our system is to provide interactive control of the viewing parameters,” lead author of the study Yuichi Taguchi, a Ph.D. student at the University of Tokyo, told PhysOrg.com. “The interactive control is essential for reproducing a dynamic 3D scene with desirable conditions, which depend on the contents of the scene, the viewer's preference, and the display specifications.”
The 3D TV system, called TransCAIP, captures a live scene using an array of 64 video cameras that are all connected via Ethernet cables to one PC, which converts images from all the video cameras into images for the display. Each video camera contains a built-in HTTP server, which sends motion JPEG sequences to the PC.
The PC then converts the 64 input views captured by the cameras to an “integral photography image” made of 60 views, which correspond to the viewing directions of the display. Using an image-based rendering technique, the PC converts the images in real-time, and then arranges the pixels to produce the integral photography image. The entire process, called light field conversion, is implemented on the single PC in real-time, requiring a few hundred milliseconds per frame.
Like all autostereoscopic displays, the new 3D TV system doesn’t require viewers to wear special glasses. Instead, the display reproduces various viewpoint images, allowing viewers to see a different image in each eye or by moving their head (the parallax effect). Although the basic principles of autostereoscopic 3D displays were developed more than a century ago, only with recent technological advances has it been possible to actually build autostereoscopic displays since they require such a large number of light rays (the resolution of a view times the number of viewpoints).
No comments:
Post a Comment