Virtual Character Definition and Animation within the MPEG-4 Standard - 3D Modeling and Animation: Synthesis and Analysis Techniques for the Human Body - page 49

of the curve shape is easily addressed. A set of control points coordinates,
weights and knots, ruling the shape of the curve, can be directly manipulated in
order to control the local curvature.
As NURBS theory is widely reported in the literature, we shall not get into details
here, however, the interested reader is referred to Piegl (1997). NURB-based
modeling is fully supported within the SMS framework. Animating a muscle
consists of updating the values of its NURBS parameters. Once the muscle
transformation is computed, it is possible to compute the new position of the skin
vertices in the muscle influence volume according to Equation (2).
Skeleton, Muscle and Skin Nodes Specification
One of the main purposes of the SMS framework is to allow the definition and
the animation of a virtual character within a hybrid 3D scene. In this context, a
scene graph architecture to describe the SMS elements is proposed. This
architecture is built according to the VRML and MPEG-4 scene graph definition
rules. The structure of the proposed architecture, therefore, relies on the
definition of scene graph nodes.
At the root of the SMS-related node hierarchy, a SBVCAnimation node is
defined. The main purpose of this node is to group together a subset of virtual
characters of the scene graph and to attach to this group an animation resource
(textual or binary). An SMS virtual character is defined as a SBSkinnedModel
node and it is related to a collection of bones, each one defined as a SBBone node,
together with a collection of muscles, defined as SBMuscle nodes. An optimal
modeling issue is addressed by defining the SBSegment node. In addition, the
SBSite node allows defining semantic regions in the space of virtual character.
The extensive description of the node interfaces is outside of the goal of this
chapter and one can find them in the MPEG-4 standard Part 16 published by ISO
(ISOIEC, 2003). Nevertheless, we briefly present the SBBone, SBMuscle and
SBSkinnedModel in order to illustrate the concepts discussed above.
The fields of the SBBone node are illustrated in Figure 11a.
The SBBone node specifies four types of information, namely: semantic data,
bone-skin influence volume, bone geometric transformation, and bone IK
constraints. Each of these components is further detailed in the following.
The SBBone node is used as a building block in order to describe the hierarchy
of the articulated virtual character by attaching one or more child objects. The
children field has the same semantic as in MPEG-4 BIFS. During the animation,
each bone can be addressed by using its identifier, boneID . This field is also
present in the animation resource (textual or binary). If two bones share the same