Masquerade: fine-scale details for head-mounted camera motion capture data

Abstract

We present Masquerade, a novel modular and expandable tool for adding fine-scale details to facial motion capture data from head-mounted cameras. After studying two important related works we developed a framework to reproduce the original approaches as well as to test equally promising alternatives. This framework has been vital for understanding the limitations of previous approaches and to explore ways to improve the results. Our final solution was a combination of algorithms and data representations that produced better results than previous works when tested with our evaluation data. Since then, Masquerade is being actively used in production for enhancing marker data with fine-scale details.