3DTV FAQ Outlines Different 3D Delivery Options
Early 3DTV broadcast services are dominated by Frame Compatible last mile delivery but a number of alternatives are being talked about. With standards to be developed and business models to be proven, this is very early days for the new format. In the run-up to IBC, Harmonic, which provides a range of headend solutions including HD encoders suited to Frame Compatible 3DTV, has provided this FAQ document to provide some more insight on this subject.
Why has 3D suddenly become such a hot topic?
The CE industry wants to trigger a TV set replacement cycle which generates higher margins at the beginning of product life cycles. At the same time major theatrical releases are in theatres internationally, stirring consumer interest. Second, the rapid increase by theatres of digital cinema technologies permits conversion of theatres to 3D with comparative ease, versus the film-based technologies of the past.
Why have the movie studios suddenly focused so heavily on 3D movies? After all, they have tried this several times before, each time being only a short fad.
The difference this time is digital cinema technologies are rapidly replacing film-based delivery systems worldwide. Beyond this, the studios spend a huge amount of money making, shipping, and maintaining film copies of each movie. The necessary standards have been published by SMPTE, the manufacturers of digital cinema equipment have ramped up production, and the studios are assisting in financing the conversion. The time, to use an old expression, is right. Once a theatre has been converted to digital cinema technology, the addition of 3D capabilities is fairly trivial and quite low cost.
What are the types of glasses technologies being used in the home?
There are three basic types of glasses currently in use in the home. The first, and most primitive, are the red/cyan anaglyph glasses. There is a slightly enhanced version that compensates for the way the eye treats red and cyan differently (called diopter glasses).
The second, with much more satisfactory performance, are circular polarized glasses. What is currently missing is standardization as to which eye gets which polarization; not a crucial issue - until you have to buy replacement glasses.
The third type are active shutter glasses, in which the display mounts a transmitter which signals the glasses (via IR or RF) to block one eye at a time in sync with the display. These glasses are totally immune to the head movement artifacts which affect polarized glasses. They are somewhat more expensive and require batteries.
How are two streams of HD pictures carried through a single transport?
Pairs of images may be “multiplexed” together in several different ways. Each method has its pros and cons. First, spatially multiplexed image formats include “Side by Side,” “Top and Bottom,” “Checkerboard,” “Line Interleave,” and “Column Interleave.” These result in an up to 50% reduction in picture resolution, but if well done are perfectly acceptable to viewers. These techniques are properly termed “Frame Compatible.”
Second, temporally multiplexed image formats termed “Time Interleave” or “Page Flip” result in only one half of the temporal rate being delivered. With movie content at 24 frames a second and 48 Hz systems being feasible, this method can be used for movies. For full motion video, however, especially sports, it would result in lowering the 60 frames per second image to 30 frames per eye. This is generally not acceptable to viewers.
What additional bit-rates do these frame-compatible methods require?
That is content dependent. For computer generated images, such as are used in animated movies, the extra bit-rate requirements are minimal. For live 3D images, however, studies have shown a bit-rate increase of up to about 35%, depending upon the number of hard edges in the picture. This is because the process of creating the frame-compatible single image increases the amount of high frequencies in the picture. High frequency data takes more bits to faithfully reproduce.
What are some of the many competing forms of image multiplexing?
One takes the left and right eye signals and just transmits one eye plus the difference between left and right eye only. This is the basis of patents pending with TD Vision. The signal for the second eye is then reconstructed at the viewer’s equipment (either in the set-top box or in the TV itself).
Advocates for this claim this is the most efficient method to enable a simulcast of 2D and 3D services. Opponents claim that this attribute is not valuable as content has to be inherently different due to the different viewing experience. Moreover, they claim that this process does not work well for occlusion (which is where you are trying to simulate an object that is only visible with one eye and being blocked from view by another near object for the other eye).
Another method subsamples both the left eye and right eye signals either vertically or horizontally to allow the combination of left and right signals to be transmitted in the same channel bandwidth as the original 2D signal. Another variant creates a checkerboard segmentation of the information. Like the horizontal or vertical subsampling above (often referred to as “side by side” and “top and bottom”) this does cause some loss of fidelity but allows the 3D signal to fit into a 2D channel bandwidth. Other systems transmit tightly synchronized but independent left-eye and right-eye views.
I have heard about the new MPEG “MVC” and “SVC” standards. Can you tell me more?
First, the acronyms: MVC stands for Multi-View Coding and is capable of compressing a potentially large number of related images (termed “views”) into one stream. It is intended for multi-camera angle services, but MPEG has retrofitted a new “Stereo Profile” aimed at 3D into the document.
SVC stands for Scalable Video Coding, and provides a way to send various “layers” of images which permit lower bit-rate, lower resolution, or lower quality images to be recovered in situations where the full quality/resolution images cannot.
The bad news related to MVC is that its efficiency may not meet the targets in all situations, especially with live video. This is to say that in some cases there will be no efficiency gain beyond “simulcasting” of the various images via AVC. The good news is that both MVC and SVC are additions to the AVC standard and documented in it as new Annexes.
Why are there two new coding schemes?
MVC and SVC are intended for different uses. MVC is an extension of AVC in a fully compatible manner. SVC extends AVC in a sometimes incompatible manner. There are sound technical reasons for this. The drawback to SVC is that a Main Profile AVC decoder may under some circumstances not necessarily be able to decode the “base layer.”
What’s all this “base” and “enhancement layer” stuff?
Both MVC and SVC are “layered” coding schemes. This means that both will generate a “base layer” and one or more “enhancement layers.” In the case of MVC Stereo Profile, the base layer will typically be the left eye view, while the enhancement layer carries the right eye view. Some have proposed carrying a frame-compatible image in the base layer and having the enhancement layer carry the portions of the picture pair which were dropped in the multiplexing process.
SVC is intended for applications in which (for example) the base layer is 720p and the enhancement carries the additional picture data to construct a 1080p image. See the question regarding “2D compatibility” below.
Now that MVC has been adopted by the Blu-ray association, will that become the dominant technology base?
What is important to the Blu-ray community is the need to supply both 2D and 3D in the same package, and they have no real constraints in bandwidth compared to operators. Operators will be delivering 3D content quite independently from any 2D counterpart and thus MVC has less to recommend it.
Is 2D compatibility really important?
What we have heard from those who actually have produced 3D content, and they have found this especially true for sports, is that the production grammar is very different between 2D and 3D productions. Camera angles which are compelling for 3D are confusing for 2D, while those widely used for 2D do not render much 3D effect. This translates into needing dual production, which adds to the cost. This factor, plus the small number of 3D viewers initially, translates into pay-per-view being the content provider’s method of choice for 3D, especially sports.
Will glasses of some form always be needed to create 3DTV?
A number of manufacturers continue to try to create auto-stereoscopic sets, with little success so far. Physics is not very favourable to this method, and thus far a big disadvantage of these sets has been the limited number of “sweet spots” for viewing. In years to come it is possible that new display technology (such as holographic projection) could overcome this, but that is purely speculative at this time. Glasses are not a big hindrance to widespread consumer adoption, however, as many see them as “cool.”
By John Moulding, Videonet
