We present a camera network system consisting of several modules of 2-3 low end cameras attached to one computer. It is not possible for a human to observe all the information coming from such a network simultaneously. Our system is designed to select a suitable viewpoint for each part of the video sequence, thus automatically creating one real-time video stream that contains the most important data. It acts as a combination of a director and a cameraman for a constrained scenario of one subject in a scene.

Creating such system involves three main subtasks - building the software and hardware platform, developing a tracking module and designing the viewpoint selection algorithm. For the tracking, which provides essential information about the position of the subject, a method based on Bayesian per-pixel classification was developed. The viewpoint selection is inspired by cinematography, which has already developed its own terminology, techniques and rules, how to make a good movie.

The resulting video can typically be used for telepresence applications or as a low-cost documentary or instruction video.