3D Scene Segmentation for Rapid Large-Scale Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional AR/VR applications face challenges in rapidly authoring and cost-effectively creating 3D presentations for large-scale, complex scenes, often requiring time-consuming 3D modeling or limited 360-degree pictures with restricted interactions.
Innovation Solution
A method and system that divide a scene into segments, using 360-degree cameras to record videos from each segment's center, allowing for low-fidelity background videos and seamless transitions to 3D models at interaction hotspots, enabling rapid authoring and interaction within large-scale industrial environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If 3D models are created for all objects in the scene, then interaction capability and visual fidelity are improved, but authoring time and cost increase significantly
Solution Approach 1:
The scene is divided into multiple segments or zones, with different levels of modeling detail applied to each segment. High-fidelity 3D models are created only for interactive objects in key segments, while other segments use lower-fidelity representations such as 360-degree images or simplified models, thereby reducing overall authoring time while maintaining interaction capability where needed.
Solution Approach 2:
Different levels of visual fidelity and interaction detail are applied locally to different parts of the scene based on their importance and interaction requirements. Critical interactive objects receive full 3D modeling treatment, while background or non-interactive elements use lighter representations, optimizing the balance between interaction capability and authoring effort.
2Productivity
If 360-degree pictures are used for navigation, then authoring speed increases, but interaction capability is limited to primitive 2D overlays
Solution Approach 1:
The system merges 360-degree image technology with 3D model integration, allowing the scene to be navigated using efficient 360-degree pictures while enabling transitions to interactive 3D models when users engage with specific objects or areas. This combination maintains high authoring speed while significantly enhancing interaction capability beyond primitive 2D overlays.
Solution Approach 2:
The system dynamically switches between 360-degree image mode and 3D model mode based on user interaction. During normal navigation, the lightweight 360-degree images provide fast rendering and authoring, while upon user engagement with interactive elements, the system transitions to detailed 3D models, providing adaptable interaction capability without sacrificing overall authoring efficiency.
3Manufacturing precision
If detailed 3D models are created for large-scale scenes, then visual fidelity is improved, but development cost and complexity increase
Solution Approach 1:
The large-scale scene is segmented into multiple smaller zones or regions, each with its own level of detail and modeling requirements. This segmentation allows the development team to focus detailed 3D modeling efforts only on specific areas requiring high visual fidelity, while other areas use simplified representations, thereby reducing overall development complexity and cost.
Solution Approach 2:
High visual fidelity through detailed 3D modeling is applied locally only to areas where it is most beneficial, such as interactive objects or focal points of interest. Surrounding or less important areas use lower-fidelity representations, optimizing the balance between visual fidelity and development complexity across the entire large-scale scene.
Data Source
AI summary
A method and system for authoring a three-dimensional (3D) presentation can involve dividing a scene into a grid of equals-sized segments or non-equally sized segments, wherein one or more 360-degree cameras record the scene from a center of each segment, and a size of the each segment is proportional to a complexity of an environment captured in the scene. One or more objects can be created within the scene for a user interaction with a 3D presentation of the scene in the context of a video stream. A logical center of the object(s) can be used as a center point for recording a 360-degree video by the one or more 360-degree cameras for use as a low fidelity background video when the user interacts with the 3D presentation of the scene. The 3D presentation can be based on a 3D model.