AR Device 3D Model Rendering via Server-Side View Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current augmented reality (AR) devices face challenges in rendering high-quality, large-capacity 3D models due to memory requirements and real-time transmission delays, making it difficult to provide seamless output of high-quality images.
Innovation Solution
A method that generates a multi-texture and 3D mesh based on a multi-view image, applying perspective correction interpolation, and transmits only the image in the view directed by the AR device's camera, using camera movement and rotation information to render and encode the image, allowing the AR device to output a high-resolution image without transmitting large-capacity data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a terminal performs rendering for a high-quality and large-capacity 3D model, then the rendering quality is improved, but a large memory is required and the execution becomes slow
Solution Approach 1:
The patent extracts only the necessary rendering task from the terminal and transfers it to a server. The terminal sends a rendering request containing 3D model data and view parameters to the server, which performs the actual rendering and returns the rendered image. This extraction resolves the memory constraint by moving the rendering workload away from the terminal's limited memory resources.
Solution Approach 2:
The patent introduces a server as an intermediary between the terminal and the 3D model data. The server acts as a mediator that receives rendering requests from the terminal, accesses the large-capacity 3D model data, performs rendering, and returns results. This intermediary resolves the contradiction by providing the terminal with access to high-quality rendering capabilities without requiring the terminal to store or process large amounts of data locally.
2Manufacturing precision
If a server performs rendering for a large-capacity 3D model and transmits the rendered image to an AR device, then the rendering quality is improved, but real-time transmission may not be performed due to the capacity of the rendered image
Solution Approach 1:
The patent applies partial action by rendering only the specific view that the AR device camera is currently directed at, rather than rendering the entire 3D model or pre-rendering all possible views. The server receives real-time camera orientation data from the AR device and renders only the necessary viewpoint, reducing the amount of data that needs to be transmitted while maintaining real-time performance.
Solution Approach 2:
The patent implements a dynamic rendering system where the server continuously adapts the rendering process to the AR device's current camera orientation and movement. The system dynamically adjusts which views to render based on real-time camera data, ensuring that only relevant images are generated and transmitted, thereby reducing transmission delays while maintaining high rendering quality.
3Manufacturing precision
If the terminal uploads a large-capacity 3D model, then the model detail is improved, but the transmission capacity and processing load increase
Solution Approach 1:
The patent extracts only the essential 3D model data needed for rendering the current view and transmits it to the server, rather than uploading the entire large-capacity 3D model to the terminal. The server performs the rendering using its own access to the detailed 3D model data, so the terminal receives only the final rendered image, not the raw model data.
Solution Approach 2:
The patent inverts the traditional approach by having the server perform rendering instead of the terminal. Instead of the terminal uploading large-capacity 3D model data and processing it locally, the terminal sends a lightweight rendering request to the server, which has access to the detailed model data and performs the rendering, then returns the result.
Data Source
AI summary
Provided is a method and device for outputting a large-capacity 3D model for an augmented reality (AR) device. A method of outputting a large-capacity 3D model for an AR device includes generating a multi-texture and a 3D mesh based on a multi-view image, generating a 3D model using the multi-texture and the 3D mesh, and transmitting, to the AR device, an image of the 3D model in a view, to which a camera of the AR device is directed, according to camera movement and rotation information of the AR device, and the AR device outputs the image in the view, to which the camera is directed.