Learning-Based 3D Model Creation with Internal Shape Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current 3D modeling techniques are limited in creating detailed, colored 3D models with internal shapes hidden by clothing or accessories, and require manual operations, which are time-consuming and costly, especially in gaming and animation industries.
Innovation Solution
A learning-based 3D model creation method and apparatus that generates multi-view feature images using supervised learning, creates a 3D mesh model with internal shape information, and produces a texture map to reconstruct both external and internal shapes of objects from input images, allowing for automatic generation of realistic 3D models without manual intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If manual 3D modeling operations are used, then detailed and accurate 3D models can be created, but labor and time costs increase significantly
Solution Approach 1:
The patent replaces manual mechanical 3D modeling operations with an automated learning-based system. The system uses neural networks and machine learning algorithms to automatically generate 3D models from input images, substituting human operators with an automated computational system that achieves both high accuracy and efficiency
Solution Approach 2:
The patent creates 3D models by copying and reconstructing objects from 2D input images. The learning-based system learns from training data to automatically generate accurate 3D representations, effectively copying the visual information from images into three-dimensional model structures without manual intervention
2Productivity
If conventional 3D reconstruction methods are used, then processing time is reduced, but the ability to reconstruct internal shapes and detailed features is lost
Solution Approach 1:
The patent segments the 3D reconstruction process into multiple specialized components: external shape reconstruction, internal shape reconstruction, and detailed feature recovery. Each segment is handled by specific neural network modules that process different aspects of the model independently, allowing simultaneous optimization of speed and accuracy for each function
Solution Approach 2:
The patent transitions from 2D image input to 3D model output by adding the depth dimension. The learning-based system infers three-dimensional spatial relationships and internal structures from two-dimensional images, effectively moving into another dimension to reconstruct complete 3D representations including hidden internal shapes
3Device complexity
If monochrome sketch images are used as input, then processing is simpler, but the output 3D model lacks color and detailed texture information
Solution Approach 1:
The patent performs preliminary colorization and texture generation during the 3D model creation process itself. Rather than adding color later, the learning-based system generates colored texture maps and detailed surface information as part of the initial model construction, preserving color and texture information from the outset
Solution Approach 2:
The patent uses texture maps as intermediary elements between the 3D mesh structure and the final colored model. These texture maps carry color, pattern, and surface detail information that is projected onto the 3D model, serving as a mediator that transfers visual information from 2D source material to 3D representation
Data Source
AI summary
Disclosed herein are a learning-based three-dimensional (3D) model creation apparatus and method. A method for operating a learning-based 3D model creation apparatus includes generating multi-view feature images using supervised learning, creating a three-dimensional (3D) mesh model using a point cloud corresponding to the multi-view feature images and a feature image representing internal shape information, generating a texture map by projecting the 3D mesh model into three viewpoint images that are input, and creating a 3D model using the texture map.