Research on Photorealistic Virtual Face Modeling

Most of the traditional face modeling methods adopt the parameter-based method to construct, but the face model constructed by this method is too smooth and ignores the detailed features of the face. To solve this problem, a virtual face modeling method based on 3D motion capture data is proposed in this paper. In order to improve the deformation method for the realistic modeling purpose, this paper divides the modeling process of personalized face into two parts: overall modification and local deformation. The overall deformation modifies the facial shape and the position of the facial features of Maya universal model. Based on the principle of radial basis function interpolation algorithm, a smooth interpolation function is constructed, and the new position coordinates of non-characteristic points are obtained by solving the linear equations, so that they are more in line with the physiological characteristics of human faces. The end result is a more realistic virtual face that you model.