Contact-Aware Motion Retargeting via Neural Network Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional motion retargeting techniques fail to accurately transfer motion data between source and target objects due to structural differences, leading to inaccuracies, inefficiencies, and self-penetrations in animated characters, which are unrealistic and distracting.
Innovation Solution
A contact-aware motion retargeting system that uses a geometry-conditioned recurrent neural network with an encoder-space optimization strategy to preserve self-contacts and prevent self-penetrations by identifying motion constraints and fine-tuning the retargeted motion to match the target object's geometry and skeleton.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional motion retargeting techniques are used to transfer motion data between objects with different skeletal structures, then motion data transfer is achieved, but accuracy deteriorates due to structural differences
Solution Approach 1:
The system transforms motion data by adjusting parameters such as joint angles, limb lengths, and skeletal dimensions to accommodate differences between source and target objects. This allows accurate motion transfer across objects with varying anatomical structures by dynamically modifying motion parameters rather than using fixed transfer rules
Solution Approach 2:
The system introduces an intermediary optimization process that acts as a mediator between source motion data and target object animation. This intermediary layer processes the motion data through iterative optimization to resolve structural mismatches, ensuring both accuracy and adaptability without direct one-to-one mapping
2Reliability
If conventional motion retargeting is applied without contact awareness, then processing speed is maintained, but visual quality deteriorates due to self-penetrations and unrealistic contacts
Solution Approach 1:
The system performs preliminary identification of contact points and self-contact regions before executing the full motion retargeting process. By pre-detecting areas where contacts occur (such as hands touching face or feet contacting ground), the system can prepare appropriate constraints and adjustments in advance, ensuring visual realism without requiring complex real-time calculations during animation playback
Solution Approach 2:
The system implements feedback mechanisms that monitor contact constraints during motion retargeting and adjust the animation accordingly. When self-penetrations or unrealistic contacts are detected, the feedback loop modifies joint angles and positioning to eliminate these artifacts, maintaining visual reliability through continuous correction
3Manufacturing precision
If motion data is transferred without preserving self-contacts, then computational efficiency is improved, but animation quality deteriorates due to loss of important body language attributes
Solution Approach 1:
The system segments the motion data processing into distinct components: contact identification, contact constraint formulation, and motion transfer optimization. By dividing the problem into manageable segments, the system can apply specialized algorithms to each segment, achieving high precision in contact preservation while maintaining overall computational efficiency through modular processing
Data Source
AI summary
One example method involves a processing device that performs operations that include receiving a request to retarget a source motion into a target object. Operations further include providing the target object to a contact-aware motion retargeting neural network trained to retarget the source motion into the target object. The contact-aware motion retargeting neural network is trained by accessing training data that includes a source object performing the source motion. The contact-aware motion retargeting neural network generates retargeted motion for the target object, based on a self-contact having a pair of input vertices. The retargeted motion is subject to motion constraints that: (i) preserve a relative location of the self-contact and (ii) prevent self-penetration of the target object.