← ArchivePaper2012
LibEE: A Multithreaded Dependency Graph for Character Animation
Martin Watt, Lawrence D. Cutler, Alex Powell, Brendan Duncan, Michael Hutchinson, Kevin Ochs
Multithreaded dependency graph architecture for character rig evaluation, enabling scalable parallel computation of complex DreamWorks rigs.
Abstract
Presents LibEE, a multithreaded dependency graph evaluation engine for character animation built from the ground up for multicore CPU architectures. The system enables concurrent evaluation of multiple graph nodes while supporting internal node parallelism, achieving over one order of magnitude speedup. Addresses production challenges including thread-safety classification, visualization tools for graph optimization, and restructuring of character systems to exploit graph-level parallelism.
How to read this
- Category
- Systems / production paper: a multithreaded rig-evaluation engine
- Contributions
- LibEE, a dependency-graph evaluation engine for character animation built for multicore CPUs
- Concurrent evaluation of multiple graph nodes plus internal node parallelism, reported as over an order of magnitude speedup
- Production practices for thread-safety classification, graph-optimization visualization, and restructuring rigs to expose graph-level parallelism
- Context
- Relates to the dependency-graph architectures underlying character rigs in animation packages, reworked for parallel evaluation in a feature-film (DreamWorks) pipeline.
- Correctness
- This is a production-engineering paper, so results are pipeline-proven on real DreamWorks rigs rather than a controlled benchmark; the reported speedup depends on rigs being restructured to expose parallelism and on correct thread-safety classification, which a reader should treat as prerequisites, not free wins.
- Clarity
- Accessible and practically framed; a first pass conveys the architecture and the engineering lessons, with deeper passes for the parallelism and thread-safety specifics.
- How to read it
- First pass for the architecture and the production lessons (thread-safety, restructuring rigs); a second pass pays off mainly if you build or parallelize a rig-evaluation system yourself.
Builds on
Nothing in the archive, this is a starting point.
Related work
- Premo: Powerful Character Rigging, Fast Animation 2023 / SIGGRAPH
- ChopRig System 2018 / Medium
- Achieving and Maintaining Real-Time Rigs 2018 / SIGGRAPH
- LibEE 2: Enabling Fast Edits and Evaluation 2018 / DigiPro
Keywords
This page summarises the entry and links to its original source. The archive never hosts or redistributes the publication itself.Show it in the full archive list →