AR Item Modeling via Spatial Scene Modeler and On-Demand Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Client devices with limited screen sizes and input/output controls face challenges in efficiently simulating items in live video feeds due to memory constraints and the need for navigating multiple user interfaces, making it difficult to model items in real-world environments effectively.
Innovation Solution
A spatial scene modeler system that uses a client device to generate a virtual frame within a live video feed, employing machine learning to categorize the environment and request relevant 3D models from a server, allowing users to select and render items within the live video feed, thereby optimizing user interface organization and reducing memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all 3D model data is stored locally on the client device, then item modeling capability is complete, but memory requirements become unmanageable
Solution Approach 1:
The patent extracts the 3D model data storage from the client device and places it on remote servers. The client device only stores minimal data such as model identifiers and metadata, while the actual 3D model files are retrieved on-demand from server storage. This extraction principle resolves the memory constraint by removing the bulk data storage requirement from the client while preserving complete item modeling capability.
Solution Approach 2:
The patent implements a universal server system that stores and manages 3D model data for multiple clients simultaneously. This server infrastructure serves multiple functions: storing models for different items, handling requests from multiple users, and providing on-demand model delivery. This multi-functionality allows the system to support complete item modeling capability across many users without each user needing local storage of all models.
2Adaptability or versatility
If multiple menu levels and user interfaces are provided for item selection, then item selection flexibility is improved, but user interface complexity increases
Solution Approach 1:
The patent segments the item selection process into distinct functional components: environment categorization (using machine learning on live video feed), model type selection, and specific item selection. This segmentation allows the interface to present only relevant options at each stage rather than overwhelming the user with all possible items simultaneously. The segmentation maintains flexibility by allowing navigation through different categories while reducing perceived complexity through progressive disclosure.
Solution Approach 2:
The patent introduces machine learning-based environment categorization as an intermediary step between the user and the 3D model selection. The system automatically analyzes the live video feed to determine the environment type (e.g., living room, bedroom, outdoor) and uses this classification to filter and organize model options. This intermediary automatically structures the interface based on context, reducing the need for manual menu navigation while preserving selection flexibility.
3Adaptability or versatility
If the client device processes and displays all available 3D models, then model selection completeness is improved, but processing time and memory consumption increase
Solution Approach 1:
The patent implements preliminary action by pre-organizing 3D models on the server according to environment categories and item types. When a user points the device at an environment, the server already has the models organized and ready for rapid retrieval based on the detected environment type. This pre-organization eliminates the need for real-time filtering and sorting operations on the client device, significantly reducing processing time while maintaining complete model selection capability.
Solution Approach 2:
The patent extracts the computationally intensive tasks of 3D model rendering and scene composition from the client device and performs them on the server. The client device only handles lightweight tasks such as capturing video feed, identifying environment categories through simple machine learning classification, and displaying the rendered result. By taking out the heavy processing requirements, the system achieves complete model selection without excessive processing time or memory consumption on the client.
Data Source
AI summary
Disclosed are various embodiments for simulating one or more virtual objects (e.g., renders) specified spatial areas of a real-world environment. Options of item models for modeling in a given spatial area can be filtered based on specified dimensions and identified features of an image of a given spatial area. A selected item model can be rendered. and continuously updated on a display device as the client device is physical moved.