Surface Normal Vector Processing for 6DoF Video Rendering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Six degree of freedom (6DoF) video rendering is computationally expensive and requires large storage and transmission capacities due to the complexity of point cloud data, making it challenging to render high-quality immersive video at high frame rates.
Innovation Solution
The implementation of a mechanism for explicit coding of surface normals within point cloud geometry and color, along with approximating surface normals from depth maps, enhances geometric reconstructions and color modeling, allowing for more efficient rendering and transmission of 6DoF video.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If point cloud data is used to represent 6DoF video, then immersive video quality is improved, but computational cost and data storage requirements increase significantly
Solution Approach 1:
The patent extracts only the essential geometric information (surface normals) from complete point cloud data. By coding surface normals explicitly within point cloud geometry and color attributes, the system retrieves directional information needed for rendering without processing entire point cloud structures, thereby reducing computational cost while maintaining geometric reconstruction quality.
Solution Approach 2:
Instead of rendering from complete point cloud data and then extracting visual properties, the patent inverts the approach by directly coding surface normals (derivatives of geometry) into the point cloud attributes during encoding. This allows the decoder to reconstruct surfaces efficiently using pre-computed normal information, reversing the traditional rendering pipeline to achieve faster playback.
2Manufacturing precision
If point cloud data is used to represent 6DoF video, then geometric detail is improved, but data transmission capacity requirements increase
Solution Approach 1:
The patent extracts and encodes only the essential geometric information (surface normals) from complete point cloud data. By representing geometric orientation through normal vectors stored in point cloud attributes rather than full coordinate information, the system maintains geometric reconstruction quality while significantly reducing the volume of data that must be transmitted.
Solution Approach 2:
The patent changes the parameter representation from full 3D point cloud coordinates to surface normal vectors encoded within point cloud attributes. This parameter transformation allows geometric information to be represented more compactly, reducing data transmission requirements while preserving the essential geometric characteristics needed for high-quality rendering.
3Productivity
If surface normals are explicitly coded in point cloud data, then rendering efficiency is improved, but encoding complexity increases
Solution Approach 1:
The patent performs preliminary computation of surface normals during the encoding phase and embeds this pre-computed information directly into point cloud attributes. By calculating and storing normal vectors in advance rather than computing them during real-time rendering, the system increases encoding complexity upfront but achieves significant rendering speed improvements during playback.
Solution Approach 2:
Instead of computing surface normals during rendering from raw point cloud geometry, the patent inverts the process by pre-computing and encoding normal information during encoding. This reversal shifts computational burden from the rendering stage to the encoding stage, improving real-time playback performance at the cost of more complex encoding operations.
Data Source
AI summary
An apparatus to facilitate processing video bit stream data is disclosed. The apparatus includes one or more processors to encode surface normals data with point cloud geometry data included in the video bit stream data for reconstruction of objects within the video bit stream data based on the surface normals data and a memory communicatively coupled to the one or more processors.