Motion Retargeting With Object Interaction for Artifact-Free Animation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Creating animations for new or custom characters or models is time-consuming and expensive due to the challenges of motion retargeting without considering environment geometry, leading to artifacts like penetration into objects or sliding.
Innovation Solution
A system utilizing a retargeting model that performs physics-based motion optimization to correct retargeted animations for different skeletons and environment geometries, incorporating machine learning to refine the process and reduce the need for new animation creation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If motion data is applied to a new or custom character skeleton using traditional retargeting methods, then the animation can be reused without creating new animations, but artifacts like penetration into objects or sliding occur due to lack of environment geometry consideration
Solution Approach 1:
The system introduces environment geometry data as an intermediary element between the motion data and the skeleton animation. This intermediary provides spatial context about objects and terrain, enabling the retargeting process to avoid penetration and sliding artifacts while maintaining high productivity through automated processing.
Solution Approach 2:
The patent replaces traditional manual animation creation and basic retargeting mechanisms with a machine learning model that automatically generates corrected retargeted animations. This substitution of the mechanical retargeting system with an intelligent system resolves the contradiction by simultaneously improving quality through physics-aware corrections and maintaining productivity through automation.
2Loss of time
If traditional retargeting is used to animate custom characters, then new animations do not need to be created from scratch, but the process is still time-consuming and expensive
Solution Approach 1:
The system performs preliminary actions by pre-processing environment geometry data and training the machine learning model on diverse motion data and skeleton configurations. This preliminary preparation enables rapid retargeting execution later, significantly reducing production time while simplifying the complexity of animating custom characters.
Solution Approach 2:
The machine learning model enables self-service retargeting by automatically correcting animation artifacts without requiring manual intervention or complex configuration. The system serves itself by learning from training data and independently generating high-quality retargeted animations, thereby reducing both time and complexity costs.
Data Source
AI summary
A system may perform animation retargeting that may allow an existing animation to be repurposed for a different skeleton and/or a different environment geometry from that associated with the existing animation. The system may input, to a machine learning (ML) retargeting model, an input animation, a target skeleton and environment geometry data of an environment for a predicted animation, wherein the ML retargeting model is configured to generate the predicted animation based on the input animation, the target skeleton and the environment geometry data of the environment for the predicted animation and receive, from the ML retargeting model, the predicted animation based on the input animation, the target skeleton and the environment geometry data of the environment for the predicted animation.