A new method for 3D face model generation for personalized game characters.
A. Tanju Erdem · IJIGS. International journal of intelligent games & simulation · 2004
A new method for generating and animating a 3-D model of a person's face is proposed. The face model is represented via a 3-D triangular mesh (geometry mesh), and a 2-D texture image. The geometry mesh is obtained by deforming a predefined standard 3-D triangular mesh based on the relative positions of the person's facial features, such as eyes, nose, ears, lips, chin, etc, obtained from the person's photographs. The texture image is obtained from the same photographs which are taken from five different particular directions, namely, front, right, left, forehead and chin, to represent the five corresponding regions of the face. The pictures are blended along region boundaries to achieve seamless color transition on the 3-D face model. A separate 3-D triangular mesh (shape mesh) is used to animate the nodes of the geometry mesh. Each triangle of the shape mesh controls the motion of a neighborhood of nodes of the geometry mesh that are connected to it. Thus, by moving the nodes of the shape mesh, which are small in number, the nodes of the geometry mesh, which can be very large in number, are animated realistically. The nodes of the shape mesh can be moved directly or indirectly. In the indirect method, the nodes of the shape mesh are moved in accordance with the so-called facial action values, which are even smaller in number than the nodes of the shape mesh.