The rules of camera tracking

As the practice of combining live-action footage and computer-generated elements becomes more common, matchmoving has taken on an increasingly valuable role in visual effects. Despite its importance, though, it’s often one of the least understood disciplines in the industry. Camera tracking in particular is a mystery to many visual effects artists, even to those who have had to do it themselves.

The basic premise of camera tracking is relatively simple. In order to get a good camera track, matchmovers must reproduce the movement of the camera that was used to film the live-action plate. If they can do this, then any 3D object that’s placed in front of their virtual camera will appear to be locked to the footage as if it were a part of the scene. However, as anyone who’s ever tried it will confirm with a knowing and rather tired look, it’s not always so simple.

The first step in this process is to ‘2D track’ features within the plate in a matchmoving program. Next, the program analyses the 2D movement of the tracks and tries to calibrate, or ‘solve’, for the camera. The software then calculates the camera’s 3D position, rotation and other settings, such as focal length. If all goes well, the virtual camera generated by the matchmoving program will closely match the movement of the original camera. The resulting camera is then exported to a 3D animation program where it will be used to render the 3D elements being added to the scene.

Over the following three pages, you’ll discover some useful camera tracking tips that will cut out a lot of time and suffering! These tips cover a wide range of topics within the matchmoving process, from shooting on greenscreen to delivering that all-important final scene. After all, there’s much more to camera tracking than just tracking cameras.