A method and system uses geometry-driven feature point analysis to
synthesize images including for example facial expressions. Given the
feature point positions (geometry) of an expression, the method
automatically synthesizes the corresponding expression image, which has
photorealistic and natural looking expression details.