Digital geometry processing for shapes in motion

Dressed 3D humans in motion

Given a motion sequence of a human in wide clothing, we estimate its body shape using a learning approach. We then use this estimation to robustly track the 3D shape over time.

Laplacian mesh sequence processing

We have developed a discrete Laplace operator for temporally coherent mesh sequences. This operator has then been used to various mesh sequence processing tasks, such as editing.

Segmentation of mesh sequences

In this project we focus on motion-based segmentation on mesh sequences. Contrary to previous works, we do not restrict to temporally coherent sequences: we deal with any mesh animation. We are able to segment a sequence according to either rigid or uniformly stretched motions.

Manifold-oriented character skinning

We proposed a new skinning framework inspired from the mathematical concept of atlas of charts: we segment a 3D model of a character into overlapping parts, each of them being anatomically meaningful (e.g., a region for each arm, for each leg, etc., with overlaps on joints); then during deformation, each vertex in an overlapping area is updated thanks to the movement of neighboring bones.

Automatic animation skeleton computation

We proposed an algorithm to automatically compute an animation skeleton, starting from a surface mesh and a few points selected by hand on the model. This considerably simplifies infographists' work, although our algorithm let them interact in various ways, if they want to.