XR Scene Graph Node Tagging for Real-Virtual Object Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing extended reality (XR) scene description technologies lack a solution to accurately distinguish between real and virtual objects in a scene, leading to inefficient resource usage and inconsistent rendering, particularly in managing collisions and lighting effects.
Innovation Solution
A method for generating an extended reality scene description that includes a scene graph linking nodes to indicate whether they correspond to real or virtual objects, using attributes like 'realNodes' and 'virtualNodes' at the scene level, and 'isRealObject' or 'isVirtual' at the node level, to facilitate efficient rendering and collision management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If scene description technologies treat all objects uniformly without distinguishing real from virtual objects, then the scene description structure remains simple, but rendering accuracy and collision detection reliability deteriorate
Solution Approach 1:
The scene description structure is segmented by introducing a distinction between real objects and virtual objects. Each object is tagged with a type identifier (real or virtual), allowing the rendering system to apply different processing rules to different object types. This segmentation enables accurate collision detection and lighting rendering while maintaining a relatively simple overall structure through the use of standardized metadata fields.
Solution Approach 2:
Different quality attributes are applied locally to real and virtual objects based on their type. Real objects receive full physical simulation and lighting calculations, while virtual objects may use simplified models. This local differentiation improves rendering accuracy for critical elements without uniformly increasing complexity across the entire scene.
2Reliability
If scene description technologies distinguish between real and virtual objects with detailed metadata, then rendering accuracy improves, but processing overhead and resource usage increase
Solution Approach 1:
The system applies partial distinction metadata to objects - only the essential type identifier (real/virtual) is stored, rather than comprehensive differentiation for all possible object attributes. This partial action provides sufficient information for reliable collision detection while avoiding the processing overhead of excessive metadata, achieving a balance between reliability and efficiency.
3Reliability
If uniform processing is applied to all objects in the scene, then processing efficiency is maintained, but lighting consistency and rendering quality deteriorate
Solution Approach 1:
The processing approach dynamically adapts based on object type. The rendering system automatically adjusts processing intensity and methods according to whether an object is real or virtual, applying full physical lighting models to real objects and optimized models to virtual objects. This dynamic processing maintains lighting consistency across the scene while preserving overall processing efficiency through intelligent resource allocation.
Data Source
AI summary
Methods, device and data stream are provided to generate, transmit and decode scene descriptions of extended reality scenes. According to the present principles, the scene graph links nodes and comprises an information indicating which nodes correspond to objects of a first list of 3D models corresponding to objects of the real environment of the scene and/or which nodes correspond to objects of a second list of 3D models corresponding to virtual objects of the scene.