I just saw an impressive video.  Andrew Jones, Ian McDowall, Hideshi Yamada, Mark Bolas, and Paul Debevec of the University of Southern California, Fakespace Labs, and Sony Corporation have created a device capable of displaying autostereoscopic (3D) still images and video to multiple viewers from multiple viewpoints at the same time.  This video shows various examples of this device as the camera moves around it.  It is quite impressive.  Their paper describes the details.