Visibility of Digital Video Artifacts in Stereoscopic 3D and Comparison to 2D

Georgia Tech has developed at testbed to explore new encoding techniques and other enhancements to stereoscopic 3DTV. A methodical subjective testing program is important in evaluating new approaches, so the testbed has been designed to explore issues such as 1) whether typical compression and network artifacts are more or less visible in stereoscopic 3DTV and under what conditions, 2) what new artifacts specific to stereoscopic 3DTV exist and what the taxonomy of these artifacts should be relative to the tolerance of humans to them, and 3) how stereoscopic 3DTV artifacts vary in visibility depending on the codec used, the bit rate, the transport method (e.g., Frame Compatible transport), and any enhancements applied to stereoscopic 3DTV. This presentation will outline the program overall, provide initial results of subjective testing using artifacted stereoscopic 3DTV video using both active shutter and passive polarization displays, and discuss next steps in the research program. — Initial subjective results from the work thus far show that 1) compared to passive polarization, the active shutter display gave a better experience in full 3D, more visibility of artifacts, and better viewing of two independent channels by multiple viewers, 2) subjects did not see a significant difference in Frame Compatible format vs. full 3D format; 3) subjects were evenly split on whether compression and network artifacts were more or less visible in 3D vs 2D, and 4) the isolation between left and right eye in both active shutter and passive polarization glasses is currently insufficient to support independent channel viewing using 3D technology.