Rob Dupre
Kingston University, UK
Vasileios Argyriou
Kingston University, UK
Download article
Published in: Proceedings of SIGRAD 2014, Visual Computing, June 12-13, 2014, Göteborg, Sweden
Linköping Electronic Conference Proceedings 106:11, p. 83-86
Published: 2014-10-30
ISBN: 978-91-7519-212-3
ISSN: 1650-3686 (print), 1650-3740 (online)
In this paper a novel approach for physics simulation prediction is proposed with applications in graphics rendering and multimedia. A prediction mechanism is introduced based on regression that aims to reduce the computational cost of simulations in a given scene by negating the need to perform physics calculations every frame. Novel features based on the energy of a scene over time are suggested for the training stage. Experiments were performed to evaluate the performance of the proposed prediction system, indicating that in cases where precision is not essential, regression tools can be utilized providing visually similar kinematics.
