Free Viewpoint Television (FTV) is a new generation of television, which builds upon currently working systems. Its biggest novelty is that the viewer can choose freely from which viewpoint he/she wants to watch the program. The viewer can change its viewpoint anytime dynamically.
To realise the system, the program should be recorded with many cameras from many different directions at the same time, from which we can even realise the view of virtual viewpoints. Many have researched the topic before and they have developed different methods of how this virtual viewpoint can be actually created from the view of the cameras.
In this paper I present the FTV service, what kind of camera system is needed for it, what kind of methods are available for creating a virtual viewpoint, what kind of coding is needed for this and finally what kind of network infrastructure is needed. I pay attention to the difficulties of realising this, then I present my new methods for optimizing the quality of the FTV service. I try out these new methods in a simulation environment, I conduct measuring and then I evaluate the results.