By Kevin Murray, NDS
Today, broadcasters are using their existing HD infrastructures to provide 3D demonstrations, and in some cases to even launch stereoscopic 3D television (S3DTV) services. The question then is whether there is in fact a need to introduce further standardisation to the set-top box to support S3DTV? Put simply, the STB (and its IDTV equivalent) is an active part of the content chain which performs a range of functions such as overlay graphics, video manipulation and trick modes. And to perform all this correctly and seamlessly, the STB requires additional information – hence the need for standardisation.
The first thing the STB needs to be aware of is the format of the video. But why? Surely it can just pass the video straight through to the display? There are two reasons. Firstly, to enable the correct format signalling to be communicated to the display using the recent HDMI extensions, enabling automatic detection – thus removing the need for the viewer to continually reach for the remote to switch between 2D and 3D modes on their television. Secondly, to allow the STB to support and correctly position overlay graphics as used by subtitles and information banners. It is important that this information is delivered in parallel as the video signal is generally manipulated by the display in a way that can render overlay graphics unwatchable. The side-by-side format used in pictures 1 to 4 shows one example of the problem. In the first image a subtitle is placed over the video without adjusting for the underlying side-by-side format. The display manipulation results in the next two images (pictures 2 and 3) that show what is seen by the left and right eyes, and an approximation of the combined result is shown in the picture 4. Clearly, if the STB is aware of the format, it can then generate the graphics so they are readable after manipulation.
Standardising a minimum set of mandatory formats, just as HDMI has done, is essential. Each extra format represents additional complexities for the STB and therefore extra costs. Should the formats not match those defined as mandatory by HDMI, format conversion may become a requirement of the STB, and whilst some format conversions are simple, others are not.
S3DTV brings the new dimension of depth, so what happens if graphics are correctly applied, but without awareness of the depth? Something quite unpleasant! There are numerous depth cues in S3DTV, but two important ones are binocular disparity – the difference between the views in the left and right eyes – and occlusion – objects obscuring others in a scene. With a simplistic graphics overlay we can produce conflicting depth cues: where binocular disparity tells us, for instance, that the graphics are behind something, but the occlusion tells us that the graphics are in front. The illustration below shows the two conflicting depth positions that the brain must resolve. This conflict destroys the stereoscopic effect and can induce headaches.
Preventing this conflict requires the provision of depth information as part of the overall broadcast data flow. Ideally, this would constitute one, or a small number of values that provide a safe depth at which graphics can be placed, and is usable for any graphic the STB needs to generate. Whilst subtitles are the most obvious use of this information other graphics can benefit too, for instance channel information banners that often appear during channel change.
And what of future developments in 3D, when we may well see improvements in the signal delivered to the viewer, such as higher resolutions, better frame rates or more than just stereoscopic views? Will they render the above areas obsolete? The same fundamental problems will exist, these broadcast signalling extensions will still be required – they will be as applicable in the long term as they are in the short term.
Kevin Murray is System Architect, New Initiatives at NDS.
Copyright Tech-i 2010. Published with kind permission of EBU Technical.