Neural Network 3D Mesh Subdivision for Geometry Retention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional digital modeling systems face challenges in accurately subdividing three-dimensional meshes while retaining detail, often resulting in overly smoothed models and requiring manual user refinement, which is time-consuming and inaccurate.
Innovation Solution
The use of neural networks to perform geometry-dependent, non-linear subdivision of three-dimensional meshes based on local geometries, determining displacement coordinates for existing and new vertices to generate a subdivided mesh that retains local geometry information without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional linear subdivision methods are used to increase the number of vertices, then the mesh resolution is improved, but the local geometry details are lost due to excessive smoothing
Solution Approach 1:
The patent applies local quality by using geometry-dependent, non-linear subdivision that adapts to local mesh characteristics. The system analyzes local geometry features (such as curvature and feature lines) and applies different subdivision strategies to different regions, preserving sharp edges and prominent features while smoothing only appropriate areas, thus maintaining local geometry accuracy while improving overall mesh resolution
Solution Approach 2:
The patent changes the subdivision parameters from fixed linear weights to dynamic non-linear weights based on local geometry analysis. By computing geometry-dependent weights that vary across different mesh regions, the system achieves adaptive smoothing that preserves local features while achieving high-resolution output
2Measurement precision
If multiple subdivision iterations are performed to achieve desired detail, then the mesh resolution is improved, but the process requires significant manual user refinement time
Solution Approach 1:
The patent implements self-service by enabling the subdivision algorithm to automatically analyze local geometry and determine appropriate smoothing weights without manual intervention. The geometry-dependent non-linear subdivision autonomously identifies feature lines and sharp edges, adjusting subdivision behavior accordingly, eliminating the need for users to manually refine meshes after each iteration
Solution Approach 2:
The patent uses feedback by continuously analyzing local geometry characteristics during the subdivision process and adjusting subdivision weights based on detected features. This closed-loop approach allows the system to automatically adapt to the mesh's geometric properties, achieving high resolution in fewer iterations without manual refinement
3Speed
If conventional linear weighted averaging is used for vertex smoothing, then the processing speed is maintained, but the mesh accuracy deteriorates due to oversmoothing
Solution Approach 1:
The patent changes the smoothing parameters from uniform linear weights to geometry-dependent non-linear weights. By computing weights based on local geometric features such as curvature and neighbor relationships, the system achieves adaptive smoothing that maintains accuracy while preserving processing efficiency through optimized weight calculation
Data Source
AI summary
Methods, systems, and non-transitory computer readable storage media are disclosed for utilizing one or more neural networks to recursively subdivide a three-dimensional mesh according to local geometries of vertices in the three-dimensional mesh. For example, the disclosed system can determine a local geometry (e.g., a one-ring neighborhood of half-flaps) for each vertex in a three-dimensional mesh. For each subdivision iteration, the disclosed system can then utilize a neural network to determine displacement coordinates for existing vertices in the three-dimensional mesh and coordinates for new vertices added to edges between the existing vertices in the three-dimensional mesh in accordance with the local geometries of the existing vertices. Furthermore, the disclosed system can generate a subdivided three-dimensional mesh based on the determined displacement coordinates for the existing vertices and the determined coordinates for the new vertices.