New dimension

A decade ago, no one could have predicted that a new generation of 3D film technology would be bringing in cinema audiences many times larger than those of their less eye-popping flat-screen variants. After the early 1950s ’golden era’ of 3D, where audiences notoriously thrilled to big studio horrors, the 1980s resurgence left bad memories of flimsy glasses, giddy headaches and plots seemingly written solely around making viewers duck from objects leaping towards them.

But modern equipment has overcome these problems. And with Hollywood’s finest directors clamouring for their slice of the action, 3D cinema is firmly back in the audience’s affections.

Now RealD, the company behind some of the biggest titles, including Monsters vs Aliens, Bolt and the concert film U2 3D, believes that the technology could be ready to make the leap from the cinema to the living room and that with the right support, 3DTV could become huge.

RealD is not alone in thinking this. Major broadcasters BBC and Sky have also demonstrated 3D TV systems and believe the technology could have a place in the mainstream within a few years.

Michael Lewis, RealD co-founder and chief executive, explained: ’I could show you a TV right now that’s great and looks fine — it is business issues that have kept it from our homes. In the same way, digital for high-def has been around for a long time, it’s just taken a long time for that technology to spread, primarilybecause people had to set business standards.’

RealD’s approach to home 3D — which still requires the use of goggles — is based on its digital projection and depth theory that is also used in other industries. The company also supplies 3D for NASA, the military and a number of automotive and aerospace companies. The reason previous 3D technology did not work and gave people headaches after 15 minutes was because on a film camera the image from the left and right projectors can never be perfectly synchronised. ’If it goes a quarter of a frame out of alignment, your brain says something’s not right here and that’s what causes people to get headaches and nausea,’ said Lewis. ’We’re able to take a single projector and install a device called Z-screen, referencing the depth axis, that slides in front of the projector.’

The Z-screen is a photo-electronic cell that alternates left and right 144 times a second, compared with film, which runs at 24 frames a second. RealD discovered that by running the frame rate really quickly, viewers had a persistence of vision that would allow them to watch 3D for a long time without headaches and eye strain. As it is running on a single projector projecting both images, it is perfectly synchronised every time, so there is no disparity between left and right.

Cinemas require a refresh rate of 2,000Hz for 3D to work effectively, and Sony digital projectors can go up to 4,000Hz. The minimum refresh rate in television is 120Hz, with some going up to 240Hz with a lot of pixels in HD for a really clear image. ’Displays are getting more detailed and the refresh is getting faster, which is moving towards supporting 3D,’ said Lewis.

RealD is now working with a number of consumer electronics companies on the development of 3D TVs, and Lewis claims TVs that support 3D could be on shelves next year. Some on the marketplace, among them Texas Instruments’ DLP, already claim to be 3D enabled now, but for the most part 3D will require a new box. Lewis believes computer and console gaming will be the big driver to get 3D TV into the home and many popular games are already produced in 3D.

True holographic technology, dubbed auto stereo, or 3D without glasses, could follow not far behind. RealD already has an early version of the technology used in digital signage, displaying an image that looks like it sticks out. It involves taking an image and cutting it up into nine different views, but would require a significant increase in processing power and the amount of pixels to get a sufficiently good image quality for TV.

As demand for 3D entertainment increases, studios are looking not only to future films — James Cameron’s Avatar is currently creating a buzz — but are eyeing their back catalogue for potential profit and to do this they need to look at the three ways in which 3D films are captured.

Live-action films are shot using two digital cameras, which makes their ’left eye and right eye’ perspectives easier to match up in post-production. Computer-generated animation automatically calculates depth to get realistic perspective, and to get the second image 3D requires, animators take a virtual camera and move it over slightly to get the perspective from the other eye. The third way is to go back into the library and use post-production techniques to generate a 3D image from a 2D picture. The information is digitised, extracted and depth is added. This has been used for The Nightmare Before Christmas and is underway for Toy Story 1 and 2 , which are coming out in 3D in October.

The technique is not restricted to animation. George Lucas is looking at creating a 3D version of the original Star Wars films, so viewers will be able to peer under the cantina table and see once and for all that Han Solo shot first (an important issue for die-hard fans).

A similar process is under development by UK firm 2D3, a division of imaging specialist the Oxford Metrics Group. The company, which notably developed the inventive Channel 4 ident, where the number 4 forms itself against a real background, uses a technique that is able to obtain depth data and realistic computer graphics using footage from a single camera.

Julian Morris, founder and chief executive, said: ’If you look at a photo of the Taj Mahal, it’s pretty obvious where you are in relation to it. But if you take a completely arbitrary scene, with nothing particularly recognisable, there’s no frame of reference. Using two cameras would give what people generally recognise as a stereo picture, but you can also calculate where the cameras are in relation to the things in the scene. Expanding this notion, with a single moving camera taking pictures from multiple positions, you can estimate the path of the camera as it took the pictures of the scene and estimate its pose [position and attitude] and from that extract 3D data.’

2D3 collaborated with Oxford University’s engineering visual geometry group run by Prof Andrew Zisserman to develop algorithms to extract depth data from images from a single moving camera. This resulted in 2D3 releasing its Boujou software package for the entertainment industry in 2000, which has been used a great deal for augmented reality, where the virtual camera is placed in the same position as the real camera to incorporate computer-generated content into a live scene.

Boujou is claimed to be the first piece of software that would take a video sequence and automatically calculate the pose of that camera and the 3D positions of features in the scene. It can only calculate the 3D position of static objects, but can automatically choose these and ignore moving objects such as cars driving or leaves blowing. Morris likens this to the way people with sight in only one eye can still judge depth sufficiently to play tennis or drive a car.

Boujou has found applications beyond the film studio and is now being used by architects to incorporate a 3D model of a new building being designed into a cityscape shot, for example, from a helicopter flying around. It can also be used to recreate realistic 3D models of objects, used widely in internet shopping to allow you to view a potential purchase from any angle using standard photos without any traditional CAD modelling.

’3D is all about replicating an image as true to life as possible and we are in the early stages of a visual revolution with holography and virtual reality as the goal,’ said Lewis. ’There are still a lot of problems to overcome, but I think this technology is as significant as the introduction of colour into film.’ It seems that the much-heralded 3D revolution could be properly underway.