Content providers should approach 3-D conversions with caution

While several industry analysts are predicting a potentially large market for 3-D TV sets around the world, one clear hurdle is the lack of content to watch at home (beyond animated “kids” movies). To solve this dilemma, consumer electronics as well as production and distribution equipment vendors have rushed to market an array of conversion boxes and software that convert 2-D content into pseudo-3-D, including adding enhanced dimension and separated planes to existing flat scenes to give the impression of 3-D.

To help give consumers more to watch, virtually all of the new 3-D TV sets from LG Electronics (opens in new tab), Panasonic (opens in new tab), Samsung (opens in new tab), Sony and Toshiba all include 2-D-to-3-D conversion circuitry that leverages new algorithms to bring depth to cooking shows and nature scenes. Still, its success often depends upon the content being viewed.

Movie studios and TV content providers are beginning to understand that the quality of such conversions often leaves viewers with a less-than-spectacular experience and can slow consumer adoption of the technology. Warnings from many in the industry have sought to make producers aware of the dangers to 3-D business models.

Director James Cameron, speaking about the current crop of movies that were originally shot for 2-D but are being converted for movie screens (and later for home delivery), has urged the TV community to embrace 3-D technology but avoid or minimize the use of 2-D-to-3-D conversion technology. At a 3-D conference in Korea, the director said, “We’re going to have 3-D TVs all around us, so we're going to need thousands of hours of sports, comedy and music and all kinds of entertainment,” but employing “cheap conversion” technology to fill the gap is not the answer. Content designed for 3-D should be shot that way, no matter how expensive it may be.

This creative idealism is butting up against the attraction among Hollywood studios for the higher-grossing 3-D movies: Six of the top 10 movies in the past year have been in 3-D. This has led film and TV producers (in some cases against their wishes) to consider 2-D-to-3-D conversion techniques as a shortcut to 3-D content. Cameron warned against such temptations saying that far from a “brilliant way to save money,” all content makers would be creating would be “eye strain and headaches.”

On the content distribution side, what we’re seeing is that “broadcast equipment manufacturers are trying to help fill the pipeline so broadcasters have something to offer their viewers, who just went out and bought a new 3-D TV set,” according to Geert-Jan Gussen, marketing manager at Axon.

The UK-based company offers its G3D100 real-time converter, which is a software suite with special features that support 3-D stereo production. This includes horizontal and vertical flip on both inputs to flip/rotate images coming from a 3-D camera set on a rig with a mirror. It can squeeze the left- and right-eye signals to an anamorphic half horizontal size and then combine them into one side-by-side stream. It also has two color correctors onboard.

A new generation of conversion technology is also now available, in software and hardware, from such companies as Autodesk (opens in new tab), Harris, JVC, Miranda Technologies, NVIDIA (opens in new tab), Panasonic, Quantel, Sony, SterGen and many others. Such conversion can be done in real time in a time-consuming frame-by-frame process or by using a key-frame-based semiautomatic method.

“All bring compromises,” said Frederik Zilly, project manager of image processing at the Fraunhofer Institute, which is developing its own device in partnership with German manufacturer Digital Vision. “Broadcasters should look at all of the available technologies and figure out which ones will work best for their purposes. Some broadcasters want to conserve bandwidth, while others will pick quality over channel count. It’s very early in the progress of 3-D TV, but if a broadcaster needs content to fill a channel, converting content you already have is the obvious choice because it’s a lot cheaper and faster to get to air. Not many will spend a lot of many in these early days when very few people can watch (3-D).”

Miranda Technologies has licensed the stereoscopic RealD format, from RealD, to broaden the conversion capabilities of its Densité 3DX-3901 stereoscopic 3-D signal processor. The RealD format is a patented version of a side-by-side 3-D formatting technology.

“By adding RealD format conversion to our 3DX-3901, the processor now offers the full breadth of stereoscopic 3-D conversion capabilities,” said Jean-Marc d’Anjou, vice president of infrastructure products at Miranda. “It can now interface between all the common 3-D formats, and thereby simplify the transition to 3-D TV playout using an existing HD infrastructure.”

As one of the first conversion boxes on the market in late 2009, JVC has sold dozens of its IF-2D3D1 stereoscopic image processors. The box works as a 2-D-to-3-D converter and as a 3-D L/R mixer for video content producers.

“With the public’s renewed interest in 3-D for both theatrically released content and beyond, many content producers are looking to repurpose existing 2-D materials to meet demand,” said Dave Walton, assistant vice president of marketing and communication at JVC. “Meanwhile, other producers creating new content need an uncomplicated way to check their 3-D footage on location. This single unit is ideal for both.”

Using proprietary algorithms developed by JVC, the IF-2D3D1 converts 2-D content to 3-D in real time, offering four 3-D mixed formats (which combine left- and right-eye images) for stereo video output on a compatible device: line by line, side by side half, above/below and checkerboard.

One example of 3-D post software that garnered a lot of attention at the recent IBC2010 was NVIDIA’s new Quadro Digital Video Pipeline suite, which, according to the company, simplifies and accelerates the production of live 3-D broadcasts. It leverages the company’s latest graphics processing units (such as the Quadro 6000 and Quadro 5000), which are based on the NVIDIA’s Fermi architecture, and it’s all integrated into a cost-effective computer workstation.

Stereoscopic 3-D images are created by processing simultaneous right- and left-eye video streams in real time. In addition, by pairing the Quadro Digital Video Pipeline with NVIDIA 3D Vision Pro active-shutter glasses, broadcast operators and directors can preview stereo 3-D content at full resolution during production.

For post production of 3-D content, professionals can edit with Adobe Premiere Pro CS5, which is also accelerated by NVIDIA Quadro GPUs. A variety of plug-ins for 3-D capture, compositing and encoding enable the tightly integrated workflow.

While there are many choices, when it comes to converting existing archives, content providers should be careful which programs they pick and how the conversion process is handled. Like “Avatar” director Cameron, industry pioneers like Jeffrey Katzenberg, CEO of DreamWorks Animation, has been equally appalled by some of the 3-D conversion work done to date.

In a keynote speech he gave earlier this year, Katzenberg said, “When it comes to 3-D conversion, here’s the bottom line: Over the years the film industry comes up with new ways to make a bad movie worse. Conversion of 2-D to 3-D is just the newest.”