January 15th, 2010 Posted in 3D Courses, Corrections and Elucidations by Mark Schubin | 
Print This Post
Print This Post
Rafe Needleman writes of his stereo blindness in cnet news and of watching Avatar in a 2D auditorium:
http://news.cnet.com/8301-19882_3-10435478-250.html
But, at home, where some viewers might want 3D and others 2D, 2D viewing might not be an option without 2D glasses.
The piece also includes this interesting sentence:
“Bruce Berkoff of the LCDTV Association and formerly a marketing executive at LG, noted that for all the hype around 3D, the television manufacturers are not really investing much in putting products on store shelves, nor are they expecting consumers to pay for it yet.”
January 15th, 2010 at 9:41 am
Since 3D content requires separate images for the left and right eye, has anyone stated the MPEG compression penalty for compressing such a stream? It seems to me that Blu-Ray or OTA HD content would necessarily either need more bitrate for equal quality or lower quality for equal bitrate vs. “traditional” 2D content.
January 15th, 2010 at 12:23 pm
It’s true that there are two views in 3-D, but it’s also true that they’re extremely similar and that bit-rate-reduction systems like MPEG’s make use of redundancy to reduce data rate. There will certainly be some overhead for the second view, but it need not be anywhere near 100%.
January 29th, 2010 at 3:14 am
Walter Gish and Christopher Vogt had a paper at the 2009 SMPTE Technology Conference discussing the coding efficiency of the MVC (multi-view video coding) extension of MPEG-4 AVC/H.264, which will be used for 3D Blu-ray coding.
They claim “the overhead for encoding stereo is between 65% and 100%, depending on
source material and the quality of the encodings.”
Which is odd, because as Mark points out about left and right eye views “they’re extremely similar”.
Gish and Vogt hypothesize that the difference between a right eye view and a left eye view (disparity) is not typically linear, but typically a more complex transformation based on change of perspective.
Unfortunately, H.264 only provides linear block motion & disparity prediction tools. This isn’t a problem for most 2D motion because most motion prediction is a short vector. However stereoscopic disparity requires much longer vectors.
If we want greater stereoscopic compression, perhaps we may need to consider solutions with more complex disparity prediction like US Patent 6144701 “An apparatus and method that applies an affine transformation to achieve steroscopic coding of video.”
January 29th, 2010 at 2:14 pm
Yes. SMPTE Director of Engineering & Standards Peter Symes mentioned that in his presentation at the Sports Tech LA 3D conference at USC on January 19. His presentation will be available soon at either this site or the Sports Video Group site.
By the way, Symes will be at the HPA Tech Retreat (along with many other 3D experts and equipment). There’s more info here: http://www.hpaonline.com