Applications of Physically Based Simulations of Elastic Rods

Schweickart, Eston

Applications of Physically Based Simulations of Elastic Rods

dc.contributor.author	Schweickart, Eston
dc.contributor.chair	Marschner, Steve R.
dc.contributor.committeeMember	James, Doug L.
dc.contributor.committeeMember	Birman, Ken
dc.contributor.committeeMember	Van Loan, Charles Francis
dc.date.accessioned	2020-06-23T17:59:21Z
dc.date.available	2021-01-17T07:00:48Z
dc.date.issued	2019-12
dc.description	114 pages
dc.description	Supplemental file(s) description: Animations and experiments described in section 4.4., High resolution images from figures 5.8 and 5.9.
dc.description.abstract	Physical simulations are widely used to recreate the motions of everyday objects, and elastic rods---long and thin deformable bodies---are an important computer animation primitive. In computer graphics, many simulation models have been developed to produce plausible animations, but many of these methods can also accomplish tasks beyond creating visualizations. This thesis demonstrates how simulation methods for animation of slender rods can be used in other applications. First, we describe physically based methods for simultaneous generation of animation and sound for deformable rods. We introduce an efficient acoustic radiation method based on dipoles, and show how to tie it to common elastic rod simulation frameworks. We present several examples of our results, including challenging scenes involving thousands of highly coupled frictional contacts. We then show how elastic rods can improve yarn geometry synthesis techniques. Prior work can generate virtual fiber curves for specific types of yarn, but does not account for how these curves deform as yarns collide. We introduce macro-fibers, elastic rods that represent groups of fibers in a yarn. We split existing yarn curves into macro-fibers and run a short relaxation simulation to allow them to rearrange locally. We show how the resulting curves can be used to predict yarn deformation in the context of knitted fabrics. By following the paths of the macro-fibers, we can generate fiber assemblies for rendering that more closely resemble the structure of actual cloth than prior work. Finally, we propose a method for tracking yarn paths through computed tomography (CT) scans of real fabrics. Existing methods are either designed for low-curvature yarn configurations, tend to fail around yarn crossings, or rely on identifying individual fibers within a pattern. Our algorithm finds ridges---chains of high-density voxels that approximate fibers---to guide centerline placement without explicitly identifying fibers, and is robust to arbitrary yarn arrangements. By reconstructing the yarn paths within these volumetric scans, we are more prepared to quantitatively validate and calibrate yarn-level cloth simulations.
dc.identifier.doi	https://doi.org/10.7298/wn3v-m255
dc.identifier.other	Schweickart_cornellgrad_0058F_11744
dc.identifier.other	http://dissertations.umi.com/cornellgrad:11744
dc.identifier.uri	https://hdl.handle.net/1813/70001
dc.language.iso	en
dc.rights	Attribution-NonCommercial-ShareAlike 4.0 International
dc.rights.uri	https://creativecommons.org/licenses/by-nc-sa/4.0/
dc.subject	Computational Geometry
dc.subject	Computer Graphics
dc.subject	Simulation
dc.subject	Sound Synthesis
dc.title	Applications of Physically Based Simulations of Elastic Rods
dc.type	dissertation or thesis
dcterms.license	https://hdl.handle.net/1813/59810
thesis.degree.discipline	Computer Science
thesis.degree.level	Doctor of Philosophy
thesis.degree.name	Ph. D., Computer Science