[by Paul Goodhead, bit-tech]
A lot of companies are pushing 3D hard right now; it’s tipped by many as the next big thing and manufacturers are positively queuing up to sell consumers anything with the word 3D in the title. However, there is one major factor holding back 3D – those silly glasses. ‘It looks great, but I don’t like the glasses’ is one of the most common comments you hear at any 3D themed event and we’re not the only ones to have noticed.
Both Sky TV and Virgin Media, the biggest two subscription TV providers in the UK, both believe that glasses-free 3DTV is what’s needed to push the technology forward and expand the market. ‘We are not a TV manufacturer, but glasses-free would be a great boost for the industry,’ says John Classy, director of 3D TV at Sky.
Kevin O’Neil, head of VOD at Virgin Media concurs, explaining that ‘3D should be a glasses-less experience in the home as soon as possible.’ We were therefore intrigued when iPont told us that it wanted to show us its proprietary glasses-free 3DTV technology at the Screen Media Expo last week.
The company’s marketing manager Tamas Motajcsek is quick to point out that glasses-free technology is nothing new; the technology to make the auto-stereoscopic screens needed to produce a glasses-free 3D image has been around for a while now, and is already available to buy on any high street in the form of the Nintendo 3DS. As a result, many of the large screen manufacturers are currently working on (or in some cases have already released) living room sized auto-stereoscopic screens already.
What’s currently missing from the glasses-free 3D mix, however, is the content, as normal stereoscopic 3D content simply doesn’t work on an auto-stereoscopic screen. This is where iPont comes in, as the company has developed an algorithm to convert traditional dual camera stereoscopic content into the multi camera auto-stereoscopic content needed for glasses-free 3D.
‘This is doing real-time video conversion from stereoscopic content to without-glasses content,’ says Motajcsek, pointing to a small form factor PC sitting below a sizeable auto-stereoscopic screen. ‘It’s taking the side-by-side stream that you get with a normal stereoscopic 3DTV service, and adding virtual cameras to create the number of views the auto-stereoscopic screen needs.’
It’s the concept of ‘views’ that makes large auto-stereoscopic screens different from their stereoscopic brethren. Anyone who has used a 3DS knows that its auto-stereoscopic effect, which is only designed to cater for a single person, only works in a very focused spot; stray out of this area and the action on screen will look blurry or disjointed. Auto-stereoscopic TVs have the same problem, but it’s one that manufacturers and content creators have found they can work around by using more cameras, or views, when filming scenes.
These extra cameras give an auto-stereoscopic screen wider viewing angles and more ‘sweet spots’ in which people can sit, as they simply provide more information from which the picture can be constructed. Producing content that works on the five, eight or nine-view screens that exist (28-view screens have even been demoed) is expensive, however, as you need a camera for every view.
Consequently, there isn’t much dedicated content around for auto-stereoscopic TVs apart from dedicated tech demos or adverts. However, Motajcsek claims this doesn’t matter too much, as converting stereoscopic 3D signals for use with auto-stereoscopic panels works well. ‘2D to 3D is a guessing game,’ says Motajcsek, ‘as the content isn’t originally made with 3D in mind, but if you already have two views (as with a standard stereoscopic feed), you can use an algorithm to create extra cameras relatively successfully.
‘We have to do this in real time, though, which is the challenge. We have only 33ms per frame to do all the calculations and conversions we need.’ It’s obvious, then, that this kind of on-the-fly translation has the potential to require a relatively hefty amount of computational power.
‘It’s done on the graphics processor rather than the CPU and requires a big GPU, was the only hint we got about what jiggery pokery was going on in the box next to us, as the company wouldn’t reveal whether it was an Nvidia or ATI GPU doing the leg-work inside the case. We did, however, manage to stick our hand behind what iPont is currently calling its 3D TV Box to confirm that the GPU was a dual-slot model that was kicking out a fair amount of heat when converting the video stream.
This appears to indicate that iPont’s dream of performing this kind of conversion on a chip inside the TV is a little way off yet, despite transistors getting continually smaller and more efficient. We also can’t help feeling that needing this kind of power on tap local to the TV is likely to be a problem for those wishing to have an auto-stereoscopic experience in their living room.
So how was the actual effect of glasses-free 3DTV? Well, right now the answer to that is that it’s ‘just about okay.’ There were three screens of various sizes and views running different content at the event, and the quality of the image on each one varied. The effect was much more subtle and immersive on the small eight-view screen than it was on the large five-view screen, but this is to be expected as artefacts and noticeable defects were larger and more obvious on the bigger screen.
This is partly because iPont still has some polishing to do on its algorithm; it’s not yet ready for market. ‘We’re still developing it,’ said Motajcsek. ‘We recognise it’s not perfect yet; we still have some work to do. I’d say we were about 95 per cent of the way there.’ Motajcsek also suggests there is some work to do for manufacturers on the hardware side too. ‘These TVs that we’re using here at this show are from some of the smaller players in the market,’ says Motajcsek. ‘We’re really waiting for some of the larger players like Sony and LG to release auto-stereoscopic screens so we can see what we can do with them.’
Regardless of the slight imperfections we saw in the test footage at the show, we were still impressed with what we saw, although the technology is more adept at providing depth to scenes than making things pop out at you. Auto-stereoscopic 3D is unlikely to be in your living room any time in the next 12 months – high panel costs and the lack of content will see to that – but it’s clear that glasses-free 3D is possible. Once you get to drop the silly shades, it’s far more likely that people will become enthusiastic about 3D.
See the original post here: http://www.bit-tech.net/hardware/2011/05/28/can-3d-shed-it-s-need-for-glasses/1