Volker Blanz of the University of Siegen and co-workers have developed software which can produce fairly convincing animations of a person's face based on a single image of that person. They map the image onto a computer model of an average human face and then use that to produce the animation. As an exercise they have animated a few old paintings: the results can be a bit Pythonesque. Speech animation of old paintings is more interesting, but his technology begins to becomes a little bit scarey when they start adding speech to video images of people. You can no longer believe anything you see on video. More details can be found here.