3D Body Model Generation via Bone-Skin Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D body model generation systems are complex and impractical for mobile devices and everyday applications, as they encode shape and pose information, making them difficult to control and adapt to new poses, especially in real-time, and do not efficiently handle bone length variability.
Innovation Solution
A bone-level skinned model is generated by disentangling bone length variability from identity-specific traits, allowing for compact models with reduced parameter counts, where bone scales can be manually specified or determined using machine learning techniques, and joint angles are used to control the pose, enabling intuitive control of the 3D body shape.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If current 3D body model generation systems encode shape and pose information together, then comprehensive body representation is achieved, but model complexity increases and real-time control becomes difficult
Solution Approach 1:
The patent segments the 3D body model into separate components: bone structure (skeleton) and soft tissue (skin). The bone model is generated first with bone scale coefficients defining the skeletal framework, then the soft tissue is draped over this skeleton. This segmentation allows independent control of bone structure and soft tissue, enabling real-time pose adaptation without retraining the entire model, thus reducing overall model complexity while maintaining adaptability.
Solution Approach 2:
The patent extracts and separates bone length variability from identity-specific traits. By isolating the bone structure as a distinct component with its own scale coefficients, the system can manipulate bone lengths independently from soft tissue characteristics. This extraction enables efficient adaptation to different poses by adjusting only bone parameters without affecting identity-specific features, resolving the contradiction between model simplicity and adaptability.
2Measurement precision
If comprehensive 3D body models are generated with high accuracy, then reconstruction precision improves, but parameter count increases making mobile deployment impractical
Solution Approach 1:
The model is segmented into bone and soft tissue components with distinct parameter sets. Bone scale coefficients (a small set of parameters) define the skeletal framework, while soft tissue is represented as a deformable mesh draped over the bone structure. This segmentation maintains high reconstruction accuracy by preserving essential anatomical relationships while dramatically reducing the total parameter count compared to full-body implicit models, enabling mobile deployment.
Solution Approach 2:
The patent applies different levels of detail and parameterization to different body regions. The bone structure uses coarse scale coefficients for efficient representation, while soft tissue uses a mesh that can be locally deformed. This local quality approach maintains high accuracy where needed (soft tissue surface details) while using compact representations for underlying structure (bone scales), achieving the balance between precision and parameter efficiency.
Data Source
AI summary
Aspects of the present disclosure involve a system and a method for performing operations comprising: receiving a plurality of bone scale coefficients each corresponding to respective bones of a skeleton model; receiving a plurality of joint angle coefficients that collectively define a pose for the skeleton model; generating the skeleton model based on the received bone scale coefficients and the received joint angle coefficients; generating a base surface based on the plurality of bone scale coefficients; generating an identity surface by deformation of the base surface; and generating the 3D body model by mapping the identity surface onto the posed skeleton model.