Description
Geometry-based Point Cloud Compression (G-PCC) is a standardized compression technology for 3D point cloud data, specified in 3GPP Release 16 and later in documents such as 26.928 and 26.998. Point clouds are collections of data points in a 3D coordinate system, representing the surface geometry of objects or scenes, often used in applications like augmented reality (AR), virtual reality (VR), and autonomous vehicles. G-PCC focuses on efficiently encoding both the geometry (spatial positions of points) and attributes (e.g., color, reflectance) of point clouds, leveraging advanced algorithms to reduce data size without significant quality loss. It operates as part of the broader MPEG-I suite for immersive media, integrating with 5G networks to enable real-time streaming and interactive experiences.
Architecturally, G-PCC employs a dual-stream approach: one stream for geometry compression and another for attribute compression. The geometry compression typically uses octree-based partitioning, where the 3D space is recursively subdivided into octants, and the occupancy of these octants is encoded using entropy coding techniques like arithmetic coding. This method efficiently represents sparse point distributions common in real-world scans. For attributes, G-PCC applies predictive coding or transform-based methods, such as Region Adaptive Hierarchical Transform (RAHT), to exploit spatial correlations among points. The compression process involves steps like quantization, where precision is adjusted to balance quality and bitrate, and entropy coding to remove statistical redundancies. Key components include the encoder, which processes raw point cloud data into compressed bitstreams, and the decoder, which reconstructs the point cloud for rendering or analysis.
How it works in practice: a point cloud captured by sensors (e.g., LiDAR or depth cameras) is input to the G-PCC encoder, which analyzes the geometry and attributes, applies partitioning and prediction, and outputs a compressed bitstream. This bitstream can be transmitted over 5G networks with reduced bandwidth, stored efficiently, or decoded in real-time on devices like VR headsets. G-PCC supports various profiles and levels tailored to different use cases, from static objects to dynamic sequences. Its role in the network is to facilitate immersive services by enabling high-quality 3D content delivery within the constraints of available bandwidth, aligning with 5G's enhanced mobile broadband (eMBB) and ultra-reliable low-latency communication (URLLC) capabilities. By standardizing compression, G-PCC ensures interoperability across devices and platforms, fostering a cohesive ecosystem for extended reality (XR) and beyond.
Purpose & Motivation
G-PCC was created to address the growing demand for efficient compression of 3D point cloud data, driven by the proliferation of immersive media and XR applications in the 5G era. Prior to its standardization, point cloud data was often uncompressed or used proprietary compression schemes, leading to excessive storage and bandwidth requirements—point clouds from sources like LiDAR scanners or 3D cameras could generate gigabytes of data per second, making real-time streaming impractical over existing networks. This limitation hindered the adoption of AR/VR, telepresence, and autonomous systems, which rely on high-fidelity 3D representations. G-PCC solves these problems by providing a standardized, high-efficiency compression method that significantly reduces data volumes while preserving perceptual quality.
Motivated by the need to integrate immersive media into 5G services, 3GPP introduced G-PCC in Release 16 as part of the broader effort to define media codecs for new use cases. Historical context includes earlier MPEG standards like video-based point cloud compression (V-PCC), which projected 3D data onto 2D planes for video encoding; G-PCC offers a more direct geometry-based approach, better suited for sparse and irregular point distributions common in real-world captures. It addresses limitations of previous approaches by enabling lower latency, better scalability, and improved compression ratios for a wide range of point cloud types, from static cultural heritage scans to dynamic automotive environments.
Furthermore, G-PCC supports the evolution towards metaverse and digital twin applications by enabling efficient transmission of complex 3D worlds. Its creation was driven by industry collaboration under 3GPP and MPEG, aiming to establish a unified standard that reduces fragmentation and promotes innovation. By solving core challenges in data handling, G-PCC paves the way for next-generation services that require rich 3D interactions, ensuring that 5G networks can deliver immersive experiences sustainably and at scale.
Key Features
- Efficient compression of 3D point cloud geometry using octree-based partitioning
- Advanced attribute compression with techniques like RAHT for color and reflectance data
- Supports both lossy and lossless compression modes for flexible quality-bitrate trade-offs
- Enables real-time streaming and low-latency decoding for interactive XR applications
- Standardized profiles for diverse use cases including static, dynamic, and LiDAR point clouds
- Integrates with 5G networks to facilitate immersive media delivery under eMBB and URLLC
Evolution Across Releases
Introduced G-PCC as a standalone compression standard for point clouds in 3GPP, defining its initial architecture with geometry and attribute compression streams. Capabilities included octree-based geometry coding, RAHT for attributes, and support for basic profiles targeting immersive media and automotive applications.
Defining Specifications
| Specification | Title |
|---|---|
| TS 26.928 | 3GPP TS 26.928 |
| TS 26.998 | 3GPP TS 26.998 |