Simulation Level of Detail for Automatic Simplification of Particle System Dynamics
	David O'Brien	obrien@cs.unc.edu
	Susan Fisher	sfisher@cs.unc.edu
	Ming C. Lin	lin@cs.unc.edu

Abstract: We present a novel framework for automatically simplifying the dynamics computation of particle systems to improve simulation speeds. Our approach is based on physically-based subdivision scheme to generate a hierarchy of approximated motion models or simulation levels of detail (SLOD). At each time step, the SLODs are updated on the fly and the appropriate SLOD is adaptively chosen to reduce the computational costs. We have developed a prototype implementation and tested it on the simulation of a water fountain and a galaxy system. The preliminary results show a significant performance gain on these scenarios with little or no loss on the visual appearance of the simulation, indicating the potential to generalize this approach to other dynamical systems.

EXAMPLE SIMULATIONS

Fountain This compares a fountain simulation without SLODs to one with SLODs. The image on the right uses SLODs and is able to maintain a rate of 30 fps while the left image noticably slows down as the number of particles increases.

Changing Region of Interest This shows one stream from the above fountain rendered as simple colored dots to help distinguish the individual clusters. The coarser Region of Interest of interest is moved and resized as the simulation runs.  Clusters automatically break up and regroup accordingly.  The wire cube shows the region of coarser detail.

Close-up of SLOD Change This shows two particle streams colliding with two cones.  The left stream and cone are all just one coarse area of simulation.  This leads to unnatural looking bounces off of the left cones.  The right cone is in a finer Region of Interest.  The large clusters are broken into smaller clusters just before bouncing off the right cone giving a more natural reaction.  Some particles are then regrouped as the move below the right cone.

Galaxy This shows the SLOD algorithm working on a much more computationally intensive n-body simulation.  This is a simulated star field of approximately 10,000 stars.  Here the speed up due to SLOD is much greater.  This video starts at a very high level of detail and gradually relaxes the constraints to increase the frame rate.

This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder. Copyright 2001.