3-D acquisition

3-D acquisition has fast become a hot topic of discussion around the coffee machines in broadcast companies, film studios and post-production facilities around the world. From broadcast production teams faced with the seemingly onerous task of covering live sporting events in 3-D to the CGI team pondering over the challenges of blending live 3-D shoots with virtual scenes, 3-D is presenting new challenges as well as a whole new set of craft possibilities for the media industry.

Ever since man has found a way of capturing a moment in time on screen, there has been a small group of determined people who have pondered the idea of capturing the same scene in 3-D. From the invention and development of film, to the introduction of television in monochrome color and HD, and now with the explosive contribution from the Internet — streaming video, YouTube and the mass community of social channels — 3-D has been there, on the sidelines, from the very beginning.

3-D is an odd concept. Of all the previous advances in image technology, it stands somewhat alone. 3-D has largely been seen as a novelty and curious technology. However, recent players in this field have come to realize that the success of 3-D will depend largely on the ability of the media industry as a whole to develop 3-D as a serious contender, to make 3-D as accessible and normal as any other of the recent developments in broadcast technology, like HD, digital cinematography or Blu-ray.

With all the latest 3-D movies and live sporting events, the methods of capturing, shooting and editing the content is a complex decision process.

3-D acquisition

There are essentially three methods of acquiring 3-D material for a production. The first is to shoot material with a 3-D camera. The second is to convert existing conventional 2D video into 3-D material, and the third is to build 3-D material in modeling software.

Shooting scenes with a 3-D camera is the most effective method of acquiring 3-D material from live subjects and scenes. This can be achieved with either a specially made integrated camera (two cameras or camcorders stuck together), or something that combines the images from two lenses onto one sensor. There have even been some experiments in designing an integrated 3-D camera with one lens.

In almost all circumstances, professionals will be shooting 3-D material using a 3-D rig. This is essentially two cameras or camcorders bolted to a piece of metalwork so that one camera captures an image for the left eye and the other for the right.

The second method of acquiring 3-D material is to convert conventional 2D material. The simplest method of achieving this is to clone-and-slip the 2D. The original 2D video can be selected as the left-eye signal, and this signal is copied (cloned) for the right-eye signal. Then the clone is slipped just a few pixels to the right. There is a small reduction in quality because only the overlap region between the two images can be used. The overall effect is to push the whole scene back into the screen. It is not really 3-D, but if used for small clips no longer than about four seconds, viewers can be fooled into thinking the whole program is 3-D.

Proper real-time 2D to 3-D conversion is a complex task. Indeed, good real-time conversion of any video scene requires a great deal of complex fast signal processing. If the original signal is used for the left-eye signal, then the right-eye signal will need to be reconstructed.

2D to 3-D converters have been available for years, but new generations are becoming more complex and sophisticated. As designers increase the overall power of image processors and understand more about shape recognition, image analysis and the probability of distance assumptions, each new generation of converter becomes better than the last.

The best converter is also the oldest — us! While live broadcast requires a real-time converter, pre-recorded material and archive footage can take its time in conversion. Each frame can be rotoscoped and analyzed automatically to save time and then tweaked manually by a team of artists. However, even taking all this into account, it is still better to shoot a scene with a 3-D camera rather than a conventional 2D camera and a converter — certainly for the time being.

The third method of creating 3-D material is by using 3-D modeling software applications. Artists use this software to create entirely virtual moving scenes complete with virtual objects, plants, creatures and people, water and other liquids, and smoke and other atmospheres.

Incidentally, this method introduces a terminology conflict. What is 3-D? Artists using this software have used the term “3-D” for a long time to describe the process of modeling virtual objects and scenes. The 3-D we are discussing should perhaps be called “stereoscopic.”

3-D rigs

3-D rig design has improved to an amazing extent in the last decade. Modern rigs are more accurate, generally lighter and more stable than older designs. Improvements continue to be made, but there is one important fact that should be remembered about 3-D rig design: There is no such thing as a perfect 3-D rig. Every rig has advantages and disadvantages, and every situation and scene requires different aspects of a 3-D rig.

On the shoot

From films such as “Avatar” and last year's 3-D World Cup, each environment has its own unique requirements on set, or in the case of location and nature shoots, out in the wild. The broadcast industry has in recent years just got to grips with 2D filming in difficult and hard to reach locations, so imagine the complex tasks that are now required when filming in those same places in 3-D.


One of the main film set requirements is quality. Although time is not limitless and there are always budgetary pressures and deadlines to meet, there tends to be more time to prepare, rehearse and retake a shot if it is not correct.

This material is also generally recorded and heavily post-produced, sometimes blended with virtual scenes or objects created in 3-D modeling software applications.

On set filming greatly benefits from a fully motorized mirror rig. This allows for the kind of narrow inter-axial spacing required to give a reasonable sense of 3-D for any scene with depths up to about 12m, while also allowing the use of large high quality prime lenses.

Fully motorized control will ensure that the rig is always accurately aligned at any focal length. This is useful for situations where zoom lenses are fitted and for situations where prime lenses are continuously changed.

Space is not a problem in this scenario, so the highest quality cameras are used irrespective of their size. The output from these two cameras can be connected directly into recorders placed nearby to the rig for immediate playback and monitoring.


Shooting in nature requires patience and timing. There tends to be time to prepare a shot properly, even though it may be difficult to find a good shooting angle. However, once recording has started, it may not be possible to reshoot. There are only certain times when animals are in the correct place and nature looks just right.

On location shoots, rigs are small, highly portable, robust and fixed. Rigs use lipstick cameras and small portable camcorders. Side-by-side rigs offer ruggedness and a compact design, but with restricted camera quality. New compact mirror rigs are being introduced all the time. However, it is unlikely that they will achieve the accuracy and ruggedness of their larger counterparts.

With issues associated with nature shoots, the two images will almost certainly need correction, especially if the cameras have zoom lenses. The most compact way of achieving this is to simply record both cameras (or use two camcorders) and make the corrections in an editing system later on. The two recordings can also be run through a 3-D correction processor.


Shooting live 3-D action at sporting events and music performances, for example, requires one main requirement: speed. Quality is still an issue, but there needs to be a clear balance between capturing a scene with reasonable quality and the need to set up equipment quickly before the event, the chance that something might be damaged during the event, and the fact that in all cases, it is impossible to reshoot a scene.

Material from this scenario is post-produced for live broadcast. Video and audio flow through production mixers (switchers) and simple effects — channel idents, score line, lyrics and song title overlays — are added on the fly, just prior to transmission. A recording is made, but only for the purposes of replay, building a highlights package and repeat transmission at a later date.

Rigs in live situations need to be remotely connected to a central studio or OB truck, which could be anything up to several kilometers from the rigs themselves. Both inter-axial and convergence adjustments are made remotely from the studio or truck. It is important to reduce cabling as much as possible, and this scenario often uses fiber cable technology to send the signals from the cameras back to the studio or truck. A popular technique is to multiplex the two cameras' HD video signals into a single 3G-SDI connection. Full intercoms, return video, and motor control are also multiplexed into this fiber.


Over the years, a number of ideas have been explored when shooting 3-D. These include cameras with varying lenses and prism arrangements that make a more portable, user friendly, single bodied camera. Ensuring that a complete workflow system is in place from beginning to end will make any 3-D shoot an effective 3-D success. We will continue to see 3-D innovation help deliver fresh and exciting experiences for viewers. BE

Paul Cameron is a trainer at Sony Professional