Transfer facial expressions with identical topology

We present a novel method to produce realistic expressions and animations by transferring existing expressions from a given facial model to a new facial model. The representation of the target model is first converted to make its topology the same as the source model. The radial basis function is employed to deform the template meshes to fit the target model. We then map all the vertices to the surface of the target model while preserving its spatial relationships with neighboring vertices. Therefore, facial expressions and motions can be faithfully transferred between models. With this method, a person can create many different models with the same animated expression even if these models have different topologies and, for any given model, a person can create many different expressions.