Cartoon Textures

Abstract

In this paper we present a method for creating novel animations from a library of existing two-dimensional cartoon data. Drawing inspiration from the idea of video textures, sequences of similar-looking cartoon data are combined into a user-directed sequence. Starting with a small amount of cartoon data, we employ a method of nonlinear dimensionality reduction to discover a lower-dimensional structure of the data. The user selects a start and end frame and the system traverses this lower-dimensional manifold to re-sequence the data into a new animation. The system can automatically detect when a new sequence has visual discontinuities and may require additional source material.