Localized Optimization for Mocap-Driven Blendshapes
Ko‐Yun Liu, Wan‐Chun Ma, Jackson Lee, Chuan‐Chang Wang, Chun‐Fa Chang · Symposium on Computer Animation · 2010
The main purpose of the proposed technique is to allow a blendshape rig to create facial expressions that are independent from the underlying blendshape poses. [JTDP03] proposed an automatic segmentation technique to divide the face into regions that have similar amount of deformation. Each region finds its best blendshape weights that conforms to the motion capture data and then propagates the weights to each vertex by radial basis functions (RBFs) [LCF00].In our case, we ask the user to provide desired facial segmentation and establish a corresponding set of motion capture data for each blendshape. The benefits are: