Along with its nihilistic cyberpunk style, the film The Matrix is famous for popularizing what’s known as “bullet time” photography. You know the shots where someone would run and jump, then they’d freeze and the camera would appear to track around them as they were frozen in mid-air? That’s bullet-time. Now, that same technology may be coming to live televised sporting events.
In traditional Matrix-style bullet-time, a series of still cameras are arranged side-by-side along a curved track, with the movie camera at one end of that track. When it’s time to freeze the action, all of the still cameras simultaneously take a shot. In post-production, the editor stays with the movie camera’s moving footage, up until the moment of the freeze. Then, they cut to a frame-by-frame sequence of the stills, as they were shot moving down the track from the movie camera. When all of those stills are run together at 24 frames-per-second, it looks like the actor has frozen, and us viewers are swinging around them.
In the current version of the new bullet time system, developed by Japanese broadcaster NHK, eight robotic sub-cameras are linked to one human-operated main video camera. As the main camera is panned, tilted and zoomed to follow the action, the sub-cameras all automatically do the same, each from its own unique viewpoint. When a freeze is required, the video source switches from the main camera to a sequence of freeze-frames captured by the sub-cameras.
As with the cinematic system, a computer is used to digitally smooth the transitions between each of the stills, making the move look more fluid. As a result, the NHK system presently isn’t able to provide bullet time shots instantaneously, but can do so within about a minute. This could make it ideal for replays.
One advantage that NHK’s system has over traditional technology is the fact that the main camera can be moving (panning, tilting or zooming) when the freeze occurs, since the sub-cameras move with it. In systems like those used in The Matrix, by contrast, all of the cameras must be locked down, requiring the action to take place in the exact area that they’re pointing at.
A demonstration of the new system can be seen in the DigInfo video below.