Automated 3D Modeling via Volume Estimation for XR Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating 3D models of objects are time-consuming, costly, and often require expertise, while automated techniques based on image processing can fail to accurately represent the object's shape due to improper feature point extraction.
Innovation Solution
A method that uses a volume estimation model trained on a set of images captured from different angles to generate a high-quality 3D model of an object, including the estimation of color values and volume density values for all positions and viewing directions, allowing for precise texture and shape representation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If traditional 3D modeling using CAD programs is performed by experts, then manufacturing precision and model quality are improved, but productivity is reduced and cost increases
Solution Approach 1:
The patent replaces manual mechanical 3D modeling operations with an automated neural network-based system. The neural network automatically estimates 3D models from 2D images without requiring expert operators to manually manipulate CAD software, thus substituting human expertise with an automated intelligent system that achieves both high quality and high efficiency
Solution Approach 2:
The system enables self-service 3D modeling where the neural network autonomously processes images and generates 3D models without human intervention. The automated processing allows any user to generate high-quality 3D models simply by providing images, eliminating the need for expert operators and significantly improving productivity
2Productivity
If automated 3D modeling based on feature point extraction is used, then productivity is improved, but manufacturing precision deteriorates due to improper feature point extraction
Solution Approach 1:
The patent changes the fundamental parameters used for 3D modeling from discrete feature points to continuous volume density distributions. Instead of extracting specific feature points that may be missed or incorrectly identified, the neural network estimates volume density at every point in 3D space, providing complete and accurate shape representation
Solution Approach 2:
The patent transitions from 2D feature point extraction to 3D volume-based modeling. By estimating volume density in three-dimensional space rather than identifying points on 2D images, the system captures the complete spatial structure of objects, ensuring accurate shape representation regardless of object features
3Adaptability or versatility
If separate software is prepared or purchased for providing 3D models on online platforms, then functionality is improved, but cost and time for implementation increase
Solution Approach 1:
The patent extracts the core 3D modeling functionality from complex dedicated software into a streamlined neural network-based system. By isolating the essential image-to-3D-model transformation capability, the system can be rapidly deployed on online platforms without requiring extensive software preparation or purchase
Solution Approach 2:
The neural network-based system provides universal 3D modeling capability that can be integrated across multiple online platforms and applications. The standardized approach works for various object types and platform requirements, eliminating the need for platform-specific software development
Data Source
AI summary
A method for providing an XR object is provided, which is executed by one or more processors, and includes receiving a plurality of images obtained by capturing an image of a target object positioned in a specific space from different directions, generating an XR object for the target object based on the plurality of images, generating a code for applying the XR object to an online platform, and transmitting the code to a first user terminal.