← ArchivePaper2021

How to Train Your Dog: Neural Enhancement of Quadruped Animations

Donal Egan, George Fletcher, Yiguo Qiao, Darren Cosker, Rachel McDonnell

MIGAcademic2 cites1 descendantMotion Synthesis ML Deformation

Neural network approach to enhance keyframed or procedural quadruped animations with learned natural motion characteristics.

Open the paper ↗Trace in the lineage graph

Abstract

Creating realistic quadruped animations is challenging. Producing realistic animations using methods such as key-framing is time consuming and requires much artistic expertise. Alternatively, motion capture methods have their own challenges (getting the animal into a studio, attaching motion capture markers, and getting the animal to put on the desired performance) and the resulting animation will still most likely require cleaning up. It would be useful if an animator could provide an initial rough animation and in return be given a corresponding high quality realistic one. To this end, we present a deep-learning approach for the automatic enhancement of quadruped animations. Given an initial animation, possibly lacking the subtle details of true quadruped motion and/or containing small errors, our results show that it is possible for a neural network to learn how to add these subtleties and correct errors to produce an enhanced animation while preserving the semantics and context of the initial animation. Our work also has potential uses in other applications, for example, its ability to be used in real-time means it could form part of a quadruped embodiment system.

How to read this

Category: Method: neural enhancement of quadruped animation
Contributions: A deep-learning approach that takes a rough quadruped animation and enhances it into a higher-quality, more realistic one
Learns to add subtle motion details and correct small errors while preserving the semantics and context of the input
Runs in real time, suggesting use within a quadruped embodiment system
Context: Relates to neural motion enhancement and quadruped animation; no prior works are listed, so it sits generally within learning-based animation cleanup and motion synthesis.
Correctness: Validated by showing enhanced animations that add detail and fix errors from rough inputs; quality of the output is bounded by the input animation and the training motion, and 'realistic' here is qualitative.
Clarity: Accessible (MIG); a first pass conveys the rough-to-refined idea, a second pass clarifies the network and data.
How to read it: Read for the problem framing (enhance rather than generate) and the real-time angle; second pass for the network details if you work on motion cleanup.

Builds on

Nothing in the archive, this is a starting point.

Built upon by

Dog Code: Human to Quadruped Embodiment Using Shared Codebooks 2024

Related work

Dog Code: Human to Quadruped Embodiment Using Shared Codebooks 2024 / MIG
Motion Retargetting based on Dilated Convolutions and Skeleton-Specific Loss Functions 2020 / CGF
ReGAIL: Toward Agile Character Control From a Single Reference Motion 2024 / MIG
Pose and Skeleton-aware Neural IK for Pose and Motion Editing 2023 / SIGGRAPH Asia

Keywords

quadruped animation neural enhancement inverse kinematics forward kinematics feed-forward neural network unit quaternions canine motion capture geodesic distance hierarchical joint angle loss gait cycle joint rotation motion retargeting mode-adaptive neural networks marker reduction embodiment system locomotion

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 →