Virtual Character Definition and Animation within the MPEG-4 Standard - 3D Modeling and Animation: Synthesis and Analysis Techniques for the Human Body - page 66

animation frameworks have been presented. A generic deformation model and
its implementation in the MPEG-4 standard through the bone and muscle
controllers has been introduced. This generic concept, in relation with dedicated
surface representation tools, like Subdivision Surfaces and M ESH G RID , recently
standardized by MPEG-4, allows hierarchical animation. It provides support for
explicitly animating only a limited set of key vertices by the designer, out of which
the animation of all other vertices is automatically calculated through an iterative
calculation scheme.
In recent years, major improvements have been reported in the field of virtual
character animation, ranging from the creation of realistic models used in cinema
movies and the development of animation production tools (e.g., motion capture
systems) to the production of on-line animations in television shows and content
streaming in distributed environments. However, research in this field is still in
an initial stage and presents challenging issues. Despite of large on-going efforts,
computer vision technologies for tracking human motion have not yet reached a
level of maturity that is satisfactory for commercial use. In order to decrease the
production cost of 3D content, re-targeting motion from motion capture data set
to different avatars is still a hot topic of research. Other important research
activities are oriented towards behavioural models for avatar and/or crowds,
building autonomous agents able to own intelligence and making virtual charac-
ters “live” with the appropriate emotional content. Even if MPEG-4 did not
explicitly analyse these issues closely, its generic framework makes extensions
possible, providing the means to an ever-evolving, living standard.
Acknowledgments
This chapter has grown out of numerous discussions with members of the
MPEG-4 AFX standardization committee and fruitful suggestions proposed by
numerous colleagues, who have provided inspiration during the germination of
the chapter. In particular, Prof. Jan Cornelis, Dr. Rudi Deklerck and Dr.
Augustin Gavrilescu from the ETRO Department of the Vrije Universiteit
Brussel have contributed to many improvements in the Humanoid animation
framework, based on the M ESH G RID representation. Some parts of the work, like
the FBA and SMS have matured thanks to assessments through projects, amongst
which the IST European Project ViSiCAST has probably contributed the most
to the quality of the outcome.