← ArchivePaper2024

Real-Time Diverse Motion In-Betweening with Space-Time Control

Data-driven framework generating diverse kinematic in-between motions with explicit user control over duration, path, and style at runtime.

Open the paper ↗Trace in the lineage graph

Abstract

In this work, we present a data-driven framework for generating diverse in-betweening motions for kinematic characters. Our approach injects dynamic conditions and explicit motion controls into the procedure of motion transitions. Notably, this integration enables a finer-grained spatial-temporal control by allowing users to impart additional conditions, such as duration, path, style, etc., into the in-betweening process. We demonstrate that our in-betweening approach can synthesize both locomotion and unstructured motions, enabling rich, versatile, and high-quality animation generation.

How to read this

Category: Method: data-driven motion in-betweening with user control
Contributions: A data-driven framework that generates diverse in-between motions for kinematic characters
Injection of dynamic conditions and explicit controls (duration, path, style) for finer spatial-temporal control
Synthesis of both locomotion and unstructured motion
Context: A controllable in-betweening method in the learned-transition lineage, building on Robust Motion In-Betweening (Harvey 2020).Builds on: Robust Motion In-Betweening
Correctness: Demonstrated on locomotion and unstructured motion with runtime controls; the abstract reports qualitative diversity and control rather than detailed quantitative comparison, so judge breadth of validation from the full paper.
Clarity: Short and approachable; a first pass conveys the controllability story well.
How to read it: First pass for what controls are exposed (duration, path, style) and how they are injected; a second pass only if you need the conditioning mechanism for real-time use.

Builds on

Robust Motion In-Betweening 2020

Built upon by

Nothing yet.

Related work

Taming Diffusion Probabilistic Models for Character Control 2024 / SIGGRAPH
WalkTheDog: Cross-Morphology Motion Alignment via Phase Manifolds 2024 / SIGGRAPH
Neural Animation Layering for Synthesizing Martial Arts Movements 2021 / SIGGRAPH
Learning Robust and Scalable Motion Matching with Lipschitz Continuity and Sparse Mixture of Experts 2023 / MIG

Keywords

motion in-betweening mixture-of-experts dynamic conditional mixture-of-experts gating network expert network autoregressive motion synthesis trajectory guidance motion matching phase manifold periodic autoencoder style control kinematic consistency loss 6d rotation representation lafan1 100style dataset spherical linear interpolation atomic trajectory onnx inference

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 →