Automatic Rigging And Cartoon Animation For Human Models

We propose a novel method for automatically generating 3D real-time human cartoon animation. Given a static human model and a generic skeleton, we adapt the skeleton to the human model and use the analogy to heat equilibrium to automatically rig the skeleton model. To create cartoon-like motion, we use Laplacian of Gaussian (LoG) filter to process the motion capture data, and obtain a more "alive" or "animated" animation style. After that, we retarget the filtered data to the rigged mesh and generate the cartoon animation sequence.To generate better results, we modify the diameter of the heat and the Euclidean distance from the vertex to the bone. The Spherical Blend Skinning (SBS) is employed to avoid collapse and candy wrapper artifacts at a low computational cost and memory space. We also use GPU to accelerate when skinning and filtering the motion data. The advantage of the presented method lies in its simplicity and real-time performance. Experiments show that our approach can produce satisfactory results and the weights vary smoothly along the surface. The system is easy to use even for novices.