Although metaballs are highly suitable to describe the human body, they are very difficult to polygonize. Thus, it is difficult to use metaballs with polygon-based models. Furthermore, because the ray method is used in rendering, it requires intensive computation and much memory. Many studies have been done to come up with a metaball polygonization method, but they all require much computation time, memory, and have little parallelism. In this paper, a new algorithm with the following three merits is proposed. First, because it makes use of a few slices of density data, it requires much less memory than the conventional methods. Second, through the use of the effective radius of a metaball, it reduces the amount of computation. Finally, by making parallel processing possible, the calculation time can be reduced as the number of nodes is increased. In addition, this paper proposes the realtime renderig hardware structure to render metaball-based model.
