Avatar Mapping Through Generic Registration for Format Interoperability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for mapping between avatars are limited to unilateral mesh representations and do not support complex avatar properties, such as semantic attributes, and lack interoperability between different avatar formats, leading to inefficiencies in animation and accessory attachment.
Innovation Solution
A generic registration schema is introduced for mapping between avatar representations, enabling motion retargeting, blendshape animation retargeting, and accessory attachment by defining correspondences between vertices, faces, skeletal joints, and semantic information using the MPEG_avatar_mapping extension in glTF 2.0 format.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a shadow mesh is used to map onto a volumetric captured mesh, then animation transfer is enabled, but the mapping is limited to unilateral mesh representations and does not support complex avatar properties
Solution Approach 1:
The patent applies universality by creating a generic registration schema that can handle multiple types of avatar properties (geometric, skeletal, semantic, accessory) within a single unified framework. The schema defines correspondence between different avatar representations enable motion retargeting, blendshape animation retargeting, and accessory attachment across various avatar formats without requiring separate mapping mechanisms for each property type.
2Adaptability or versatility
If avatar representations are mapped using existing mesh linking techniques, then animation transfer works, but interoperability between different avatar formats is not supported
Solution Approach 1:
The patent introduces a generic registration schema as an intermediary layer that mediates between different avatar representations. This schema acts as a universal translator that can convert between various avatar formats (e.g., volumetric captured mesh, shadow mesh, skeletal avatar) while maintaining mapping accuracy. The schema provides standardized correspondence definitions that enable reliable interoperability without sacrificing precision.
3Adaptability or versatility
If complex avatar properties and semantic attributes are added to avatar representations, then avatar functionality is enhanced, but mapping and registration between different representations becomes more difficult
Solution Approach 1:
The patent applies segmentation by dividing the avatar representation into distinct property types (geometric properties, skeletal properties, semantic attributes, accessory information). The generic registration schema then handles each property type separately through dedicated correspondence mechanisms, making the overall registration process more manageable. This modular approach allows complex avatar functionality to be maintained while simplifying the registration process through structured property organization.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The system, method and device described herein is directed to the registration/mapping between different avatar representations including shape (e.g., vertices and/or faces, etc.) and semantics (e.g., skeletal joints, shape annotation). The system, method and device include a generic schema to map two different avatar representations. Such a mapping may allow, for example, motion retargeting, blendshape animation retargeting, semantical transfer of body parts annotations and/or attachment points of accessories.