Methods to compress simulation data are invaluable as they facilitate efficient transmission along the visual effects pipeline, fast and efficient replay of simulations for visualization and enable storage of scientific data. However, all current approaches to compressing simulation data require access to the entire dynamic simulation, leading to large memory requirements and additional computational burden. In this paper we perform compression of contact-dominated, rigid body simulations in an online, error-bounded fashion. This has the advantage of requiring access to only a narrow window of simulation data at a time while still achieving good agreement with the original simulation. Our approach is simulator agnostic allowing us to compress data from a variety of sources. We demonstrate the efficacy of our algorithm by compressing contact-dominated rigid body simulations from a number of sources, achieving compression rates of up to 360 times over raw data size.
@article{Jeruzalski:CAOC:2017,
title = {Collision-Aware and Online Compression of Rigid Body Simulations via Integrated Error Minimization},
author = {Timothy Jeruzalski and John Kanji and Alec Jacobson and David I.W. Levin},
year = {2018},
journal = {Computer Graphics Forum (Proc. SCA)},
}
This work is funded in part by NSERC Discovery Grants (RGPIN2017–05235 & RGPAS–2017–507938 & RGPIN–2017–05524, RGPAS–2017–507909), Connaught Funds NR2016–17, the Canada Research Chairs Program, and a gift by Adobe Systems Inc.