Skip to main navigation Skip to search Skip to main content

Semantic-Driven Multi-character Multi-motion 3D Animation Generation

  • Hui Liang*
  • , Fan Xu
  • , Junjun Pan
  • , Zhaolin Zhang
  • *Corresponding author for this work
  • Zhengzhou University of Light Industry

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Semantic-driven animation generation significantly eases the workload of animators, but it still confronts a variety of challenges (e.g., natural language input, temporal reasoning under visualization, linking natural language with graphical systems, etc.). This paper tackles the issue of synchronized motion juxtaposition in terms of temporal visualization. We utilize semantic dependency analysis, prior probabilities and the Semantic Action Graph to extract and fuse synchronous motions, thus creating an advanced system for generating Semantic-driven 3D animations. The results demonstrate that our system proficiently generates natural and coherent 3D animations from text descriptions involving multiple characters and actions. This effectively overcomes the synchronized motion juxtaposition challenge and advances the field of Semantic-driven animation creation.

Original languageEnglish
Title of host publicationComputer Animation and Social Agents - 37th International Conference, CASA 2024, Revised Selected Papers
EditorsNadia Magnenat Thalmann, Xinrong Hu, Bin Sheng, Daniel Thalmann, Tao Peng, Weiliang Meng, Jin Huang, Lei Zhu, Xiong Wei
PublisherSpringer Science and Business Media Deutschland GmbH
Pages268-280
Number of pages13
ISBN (Print)9789819626809
DOIs
StatePublished - 2025
Event37th International Conference on Computer Animation and Social Agents, CASA 2024 - Wuhan, China
Duration: 5 Jun 20247 Jun 2024

Publication series

NameCommunications in Computer and Information Science
Volume2374 CCIS
ISSN (Print)1865-0929
ISSN (Electronic)1865-0937

Conference

Conference37th International Conference on Computer Animation and Social Agents, CASA 2024
Country/TerritoryChina
CityWuhan
Period5/06/247/06/24

Keywords

  • animation generation
  • semantics
  • synchronized motion
  • synchronized motion juxtaposition challenge

Fingerprint

Dive into the research topics of 'Semantic-Driven Multi-character Multi-motion 3D Animation Generation'. Together they form a unique fingerprint.

Cite this