← ArchivePaper2008

Real-Time Motion Retargeting to Highly Varied User-Created Morphologies

Chris Hecker, Bernd Raabe, Ryan W. Enslow, John DeWeese, Jordan Maynard, Kees van Prooijen

SIGGRAPHIndustrial211 cites3 descendantsRetargeting

Real-time motion retargeting for highly varied character body shapes, enabling game characters with diverse proportions to share animation.

Open the paper ↗Trace in the lineage graph

Abstract

This paper presents a system for animating characters whose skeleton morphologies are unknown when the animation is authored, developed for the game Spore where players create their own creatures. Animators author motion in a tool called Spasm by attaching semantic information, contexts and movement modes, that records keyframed motion in a generalized, character-independent form. At runtime the generalized data is specialized onto specific player-created characters to produce pose goals that are fed to a robust Particle IK solver, with stylized locomotion synthesized for arbitrary leg configurations and passive secondary animation added for appeal. The IK solver treats the skeleton as particles linked by length constraints and uses a two-phase spine-then-limb solve tuned for natural poses under conflicting goals.

How to read this

Category: Method: real-time motion retargeting to unknown morphologies
Contributions: Animates characters whose skeleton morphology is unknown at authoring time, as needed for player-created creatures
An authoring tool where animators attach semantic information (contexts and movement modes) to record motion in a character-independent form
Runtime specialization onto arbitrary skeletons via a robust particle IK solver, with synthesized stylized locomotion for arbitrary leg configurations and passive secondary animation
Context: Extends the motion-retargeting line begun by Retargeting Motion to New Characters (Gleicher) to the extreme case of arbitrary, user-authored morphologies in a shipped game.Builds on: Retargeting Motion to New Characters
Correctness: Presented as a shipping production system for highly varied creatures, with a two-phase spine-then-limb particle IK solve tuned for natural poses under conflicting goals; readers should keep in mind it favors robustness and appeal over biomechanical accuracy, and trade-offs are tuned to the game's needs rather than measured against ground truth.
Clarity: Very readable and practical; a first pass conveys the generalize-then-specialize approach and a single careful pass largely suffices.
How to read it: Read it as a systems-design case study: focus on the semantic authoring representation and the particle IK two-phase solve; a second pass is worth it for the IK tuning if you build retargeting for variable rigs.

Builds on

Retargeting Motion to New Characters 1998

Built upon by

Related work

Normalized Euclidean Distance Matrices for Human Motion Retargeting 2017 / MIG
Dog Code: Human to Quadruped Embodiment Using Shared Codebooks 2024 / MIG
Motion Retargeting for Crowd Simulation 2015 / DigiPro
Motion Warping 1995 / SIGGRAPH

Keywords

motion retargeting inverse kinematics particle ik solver user-generated content character morphology procedural animation semantic specification generalization and specialization locomotion synthesis gait system spore secondary animation length constraints spine splines variants keyframe animation

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 →