Photographic or videographic data are either used by themselves for facial animation or, more frequently, combined with other model data such as when a face is texture mapped onto a head shaped surface. These techniques may be two dimensional or three dimensional, with stereo views and processing.
Ultimately the photographic or video image must be reduced to a digital form. However, for some purposes a standard video image may lack sufficient detail.
Even with a single photographic image there are problems of registering of the image with appropriate points on the model. These are most often solved by hand work. Some 3-D scanners take video images simultaneous with rangefinding and produce a 3-D model with appropriate pixels associated with each portion of the surface. The Cyberware scanners are examples of this type. If multiple images of the same face are photographed and one wishes to use these in an animation the faces must be carefully registered one with the other and be of the same size else artificial movements will be introduced or the entire head could appear to pulsate. The manually located fixed point for a series of facial expressions or for a series of visemes (visual speech elements like phoneme) is normally the center of the bridge of the nose. This is the point that should be in registration to avoid artifacts. Nostril tracking has been used successfully for automatic registration and tracking of the mouth area . Even if the photographs taken are lighted consistently, variations in processing and more importantly in the flat bed scanners will cause variations in intensity and color among the scanned images. These must be matched for color and brightness or they will cause even more annoying artifacts if several facial images are used. Also the images must be lighted consistently, particularly if the subject moves. One can introduce shadows that appear to jump with a change of facial expression.
When stereo imaging is used special lighting is often employed. This will be discussed in Section . Special lighting techniques are applied in computer lipreading systems in order to make the oral cavity have sufficient light and shadow. For these applications the camera to subject position has been fixed with a harness .
Another photographic technique used in producing animations (rather than data gathering) is rotoscoping. A rotoscope is basically a combination of a movie projector and a movie camera. An image is projected on an animation table which then guides the production of a modified drawing that is subsequently filmed and is in registration with the projected image. This is a very old technique used first in production of ``Betty Boop'' cartoons.