The tools described, permit including large numbers of complex dynamic models in a VRML world easily and efficiently while maintaining high frame rates. We describe three tools that together provide an environment for authoring cullable, dynamic, rigid-body objects in VRML and Java: a code transformation tool that exploits approximations to dynamical systems to enable culling; a runtime layer generator, which defines a simple standard interface between a VRML browser and dynamical systems described in Java; a rigid-body modeler, which allows users to interactively design the runtime layer and preview the dynamic behavior. The article describes these tools, including some example systems, and discusses the runtime performance improvements obtained. Our tools are applicable if the spatial range of the dynamic model can be bounded by a static volume, the model is closed to outside influence, the underlying equations are continuous, and the dimension (number of degrees of freedom) of the system is small. Note that while the article focuses on VRML and Java as the target environment, the underlying techniques apply to any rendering and language environment.
[1]
Heekuck Oh,et al.
Neural Networks for Pattern Recognition
,
1993,
Adv. Comput..
[2]
David A. Forsyth,et al.
Efficient dynamics modeling for VRML and Java
,
1998,
VRML '98.
[3]
C. S. Hsu,et al.
Cell-to-Cell Mapping
,
1987
.
[4]
P. Holmes,et al.
Nonlinear Oscillations, Dynamical Systems, and Bifurcations of Vector Fields
,
1983,
Applied Mathematical Sciences.
[5]
C. Hsu,et al.
Cell-To-Cell Mapping A Method of Global Analysis for Nonlinear Systems
,
1987
.
[6]
David A. Forsyth,et al.
View-dependent culling of dynamic systems in virtual environments
,
1997,
SI3D.
[7]
Geoffrey E. Hinton,et al.
NeuroAnimator: fast neural network emulation and control of physics-based models
,
1998,
SIGGRAPH.
[8]
Jessica K. Hodgins,et al.
Simulation Levels of Detail for Real-time Animation
,
1997,
Graphics Interface.