GPU-based Fast Motion Synthesis of Large Crowds Using Adaptive Multi-Joint Models
- Authors
- Sung, Mankyu; Kim, Yejin
- Issue Date
- 22-Mar-2019
- Publisher
- MDPI
- Keywords
- crowd simulation; motion synthesis; multi-joints model; character animation; GPU acceleration
- Citation
- SYMMETRY-BASEL, v.11, no.3
- Journal Title
- SYMMETRY-BASEL
- Volume
- 11
- Number
- 3
- URI
- https://scholarworks.bwise.kr/hongik/handle/2020.sw.hongik/1831
- DOI
- 10.3390/sym11030422
- ISSN
- 2073-8994
- Abstract
- This paper introduces a GPU (graphics processing unit)-based fast motion synthesis algorithm for a large crowd. The main parts of the algorithms were selecting the most appropriate joint model given adaptive screen-space occupancy of each character and synthesizing motions for the joint model with one or two input motion capture data. The different joint models had a character range from fine-detailed and fully-articulated ones to the most simplified ones. The motion synthesizer, running on a GPU, performed a series of motion blending for each joint of the characters in parallel. For better performance of the motion synthesizer, the GPU maintained a novel cache structure for given speed parameters. Using the high computation power of GPUs, the motion synthesizer could generate arbitrary speeds and orientations for the motions of a vast number of characters. Experiments showed that the proposed algorithm could animate more than 5000 characters in real-time on modest graphics acceleration cards.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - School of Games > Game Software Major > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.