Abstract
Lip model plays an important role in virtual facial expression and speech simulation. However, the structures and movements of existing lip models are very simple, and the common motions such as protruding, pouting and sipping are failed to totally realize. In order to solve these problems, this paper presents a new double-layer muscle-controlled lip model. This method adopts the approach from anatomy to classify lip muscles into deep and superficial ones. The deep muscle is employed to control the corners of the mouth position and the size of mouth fissure. The superficial muscle is treated as a new arc muscle, and it uses catenary function to fix the parameters under different shrinkages and fit the process of protruding and sipping etc. The experiments show that the method can control the model flexibly and simulate some lip movements that accord with physiological motion characteristics.