Coding of dynamic texture for mapping on 3-D scenes

As the availability of powerful 3-D scene renderers grows, the demand for h igh visual quality 3-D scenes is increasing. Besides more detailed geometri c and texture information, this presupposes the ability to map dynamic text ures, This is obviously needed to model movies, computers, and TV screens b ut also, for example, for the landscape as seen from inside a moving vehicl e or shadow and lighting effects that are not modeled separately. Downloading the complete scene to the user, before letting him interact wit h the scene, becomes very impractical and inefficient with huge scenes. If, as is often the case, a back channel is available! on-demand downloading a llows the user to start interacting with the scene immediately. Specifically for dynamic texture, if we know the viewpoint of the user (or several users), we can code the texture taking into account the viewing con ditions, i.e., coding and transmitting each part of the texture with the re quired resolution only. Applications that mould benefit from view-dependent coding of dynamic textures include (but are not limited to) multiplayer th ree-dimensional (3-D) games, walkthroughs of dynamic constructions or scene s, and 3-D simulations of dynamic systems. In this paper, the feasibility of such a scheme based on an adapted optimal level allocation video codec is shown together with the huge data-rate red uctions that can be achieved with it.