Tuning facial animation in a mocap pipeline

Abstract

In a practical mocap pipeline, it is frequently necessary for artists to manually adjust the results of the automatic retargeting. This is due to various reasons, including errors, late changes in artistic requirements, and the fact that subtle aspects of the original performance may be lost in the retargeting process. However, facial animation editing takes a lot of time. Once automatic retargeting is done, every frame in the shot becomes a keyframe, making temporally coherent editing challenging. In addition, the blendshape weights resulting from retargeting can be quite different from what artists usually use during their manual keyframing [Seol et al. 2011].