Volumetric Video Compression via Depth-Based Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face significant challenges in compressing and transmitting high-quality live-action three-dimensional volumetric video data, which generates extremely large datasets due to the combination of stereoscopic camera and LIDAR data, making it difficult for consumer-grade systems to store and stream effectively.
Innovation Solution
A compression and decompression algorithm that categorizes data into 'close', 'intermediate', and 'distant' regions, using fully rendered geometric shapes for close objects, two-dimensional projections for intermediate objects, and skybox projections for distant objects, significantly reducing data complexity while maintaining visual fidelity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-quality live-action three-dimensional volumetric video data is captured using stereoscopic cameras and LIDAR, then visual fidelity and 3D rendering quality are improved, but data size and complexity increase dramatically
Solution Approach 1:
The patent segments the three-dimensional space into multiple depth regions (foreground, midground, background) and applies different compression techniques to each region. This allows high-quality preservation for close objects while using more efficient compression for distant objects, thereby reducing overall data size while maintaining visual fidelity where it matters most.
Solution Approach 2:
The patent applies local quality by differentiating compression quality across different spatial regions. High-quality rendering is applied to close objects in the foreground, while lower-quality but still acceptable representations are used for distant objects. This local differentiation optimizes the balance between visual quality and data efficiency.
2Quantity of substance
If data is compressed to reduce size and complexity, then storage and transmission feasibility are improved, but visual quality and rendering accuracy deteriorate
Solution Approach 1:
By segmenting the scene into depth-based regions, the patent can apply appropriate compression levels to each segment. This ensures that visual quality is maintained for regions that require it (close objects) while achieving significant compression for regions where extreme detail is less critical (distant objects).
Solution Approach 2:
The patent applies partial compression rather than uniform compression. It uses stronger compression for distant objects where the human visual system is less sensitive to details, while using weaker compression for close objects. This partial action approach optimizes the trade-off between data size reduction and visual quality preservation.
3Productivity
If traditional compression algorithms are applied to volumetric video data, then processing speed is maintained, but compression efficiency and data reduction are insufficient
Solution Approach 1:
The patent segments the processing task by depth regions, allowing different compression algorithms and parameters to be applied to each segment. This enables optimized compression efficiency for each region while maintaining overall processing speed through parallel processing of different depth layers.
Solution Approach 2:
The patent changes compression parameters dynamically based on depth distance. Compression strength, resolution, and other parameters are adjusted according to the depth region being processed, enabling more efficient data reduction in distant regions while maintaining quality in close regions.
4Ease of manufacture
If consumer-grade systems are used for storage and transmission, then system accessibility and cost-effectiveness are improved, but the ability to handle large volumetric video datasets is insufficient
Solution Approach 1:
By segmenting the data into depth regions and applying differential compression, the patent reduces the overall data size to levels that consumer-grade storage systems can handle. This makes the technology accessible to consumers without requiring expensive professional-grade storage infrastructure.
Solution Approach 2:
The patent changes data representation parameters (compression level, resolution) based on depth and viewing requirements, producing data sizes that are compatible with consumer-grade storage and transmission capabilities while still providing immersive 3D experience.
Data Source
AI summary
A method for compressing geometric data and video is disclosed. The method includes receiving video and associated geometric data for a physical location, generating a background video from the video, and generating background geometric data for the geometric data outside of a predetermined distance from a capture point for the video as a skybox sphere at a non-parallax distance. The method further includes generating a geometric shape for a first detected object within the predetermined distance from the capture point from the geometric data, generating shape textures for the geometric shape from the video, and encoding the background video and shape textures as compressed video along with the geometric shape and the background geometric data as encoded volumetric video.