Re-using Traditional Animation: Methods for Semi-Automatic Segmentation and Inbetweening

Abstract

A large body of traditional animation exists that contains characters with poses, expressions, and appeal not easily achievable with modern 3D techniques. To create new uses for this body of animation, this paper presents components of a system that can help incorporate the animation into re-usable libraries. In particular, we discuss two semi-automatic techniques that allow the re-use of traditional animation. First, support vector machines are used to segment cartoon images from their backgrounds for incorporation into an image library, for such applications as re-sequencing. Second, a radial basis function implicit surface modeling technique and a fast nonrigid elastic registration algorithm provide inbetween contours and textures given two key images of traditional animation. Our system is fast, model-free, and requires minimal animator intervention.