Luxomatic: Computer Vision for Puppeteering

MIT-TR-512 |

Published by Microsoft

MIT Media Laboratory Perceptual Computing Section Technical Report

In this tech note we describe the Luxomatic performance animation system, a computer vision system that tracks a puppeteer’s hand to map hand position and shape to an animated graphical model. An eigenvector decomposition of typical hand images are used to both track the hand and extract two-dimensional pose information in real time. Luxomatic was constructed as a class project in 1997.