Game Development Reference
In-Depth Information
high complexity mode, a more detailed face model is adapted, using eye and
mouth features, eyebrow and nose features, and chin and cheek contours.
Experimental results with natural videophone sequences show that with this
framework automatic 3D face model adaptation with high accuracy is
possible.
Introduction
In the last few years, virtual humans and especially animated virtual faces (also
called talking heads) have achieved more and more attention and are used in
various applications. In modern computer games, virtual humans act as football
players or Kung Fu fighters. In movies, highly realistic animated virtual humans
are replacing real actors (e.g., in the science fiction movie “Final Fantasy”). On
the Internet, animated virtual faces are acting as news announcers or sales
agents. In visual communication applications, like video telephony or video
conferencing, the real faces of the participants are represented by virtual face
clones of themselves. If we take a closer look at the technology behind these
animated faces, the underlying shape of a virtual face is often built from a 3D
wireframe consisting of vertices and triangles. This wireframe is textured using
textures from a real person's facial image. Synthetic facial expressions are
generated by animating the 3D wireframe. Usually, the face is animated by
movement of the wireframe's vertices. In order to produce natural looking facial
movements, an underlying animation structure (providing rules for animation) is
needed, simulating the behavior of a real human face.
The creation of such an animated face requires generating a well-shaped and
textured 3D wire-frame of a human face, as well as providing rules for animation
of this specific 3D wireframe. There are different ways to create an animated
face. One possibility is that an animated face is created manually by an
experienced 3D modeler or animator. However, an automatic approach is less
time consuming and is required for some applications. Dependent on the specific
application and its requirements, different ways for the automatic creation of an
animated face exist.
For 3D modeling of the shape of the head or face, i.e., for generation of the 3D
wire-frame, techniques that are common for the 3D modeling of objects in
general could be used. With a 3D scanner, a laser beam is sent out and reflected
by the object's surface. Range data from the object can be obtained and used for
3D modeling. Other approaches use range data from multi-view images (Niem,
1994) obtained by multiple cameras for 3D modeling. All these techniques allow
a very accurate 3D modeling of an object, i.e., a human head or face. However,
Search WWH ::
Custom Search