Computer Animation and Virtual Worlds

In this work, we show an unsupervised method to extract a latent semantic layer based on geometric features, which provides a new way to represent and understand human motion capture data. This new semantic layer is composed by the combination patterns of mo-words which frequently apper in motions.

We propose an approximate mode for the generation of bubbles in water–solid interaction on the fly, which takes into account the influence of gas concentration, solid material, and water–solid velocity difference. As the surface air particle of bubbles is treated as a virtual nucleation site, bubbles absorb air from water and grow. The two-way coupling of bubbles with water and solid is calculated by a drag force, so the bubbles’ flow on the solid surface and deformation in the movement process can be simulated.

This paper proposes a novel method for a real-time cutting simulation of deformable objects using meshless method. Topological relations among simulation nodes are defined as an undirected graph and refined by removing the edges of the graph that is intersected by the cut surface. Our approach utilizes a bounding volume hierarchy (BVH) to accelerate the computation of the intersection test and provides efficient reconstruction algorithm on the basis of the two-level BVH structure.

This paper proposes a new marker design for the treatment of occlusions. It uses customizable textures that let users design their own markers or expand the marketing opportunities of the companies. Moreover, multiple tracking techniques are combined and executed in real time in order to achieve a robust and fast tracking. Additionally, the new marker design offers new interaction possibilities that can be used to develop new applications.

Existing deformation techniques are oblivious to salient structures that often capture the essence of 3D meshes. Combining with gradient domain technique, we propose an alternative approach to preserve these structures in the volumetric subspace. Massive experimental data demonstrate the effectiveness and efficiency of our algorithm.