It is a super hot topic nowdays, in computer graphics industry ,that producing realistic, expressive animated faces. This technology has been used in creating believable virtual characters for moives and television.(e.g The Curious Case of Benjamin Button )
I am quite interested in the topic of “making face” . As it seems so COOL and FUN!!
Making Faces Guenter, Grimm, Wood, Malvar, Pighin 1997
Aim : Take video of an person, and generate a representation that allows reconstruction from any view
Take video with 6 cameras of a person speaking with 182 colored dots glued all over their face
Track correspondences between dots(camera to camera & frame to frame) to determine dot motion over time
Generate a 3D model by scanning the head (with the dots)
For each frame, compute model vertex motion according to the (known) locations of the dots
Two stage fit: first determine motion for a grid of points, then determine vertex motion
To render, simply texture map the 3D fitted model
Textures come from the video streams
Remove dots from the textures, and the resulting holes filled in
Blend textures into one view independent texture
A Morphable Model For The Synthesis Of 3D Faces Blanz , Vetter 1999
The technique derives a morphable face model by transforming the shape and texture of the examples into a vector space representation. New faces and expressions can be modeled by forming linear combinations of the prototypes.
An interesting topic and What I will be working on in future: - exchange face in images...
It is also based on the the Morphable Model, which captures the range of possible shapes. and textures observed in a dataset of 3D scans.
This system applys an algorithm that estimates a textured 3D face model from a single image, along with all relevant scene parameters, such as 3D orientation, position, focal length of the camera, and the direction and intensity of illumination .
Given a set of about 7 feature points that are manually defined by the user in an interactive interface, the algorithm automatically fits a Morphable Model of 3D faces to the image and optimizes all model parameters.