NVIDIA announces the implementation of Multi-View High Efficiency Video Coding (MV-HEVC) encoder in the latest NVIDIA Video Codec SDK release, version 13.0. This significant update marks a major leap forward in hardware-accelerated, multi-view video compression. It offers enhanced compression efficiency and quality for stereoscopic and 3D video applications as compared to simulcast encoding.
Understanding MV-HEVC
MV-HEVC is an extension of the High Efficiency Video Coding (HEVC) standard, designed to efficiently compress multiple video views of the same scene captured from different vantage points. It addresses the limitations of traditional video-coding methods such as simulcast encoding, which often resulted in high bitrates for multi-view content and lacked efficient inter-view prediction.
Key features in NVIDIA Video Codec SDK 13.0 MV-HEVC
The MV-HEVC implementation in the Video Codek SDK 13.0 includes the following features:
- Support for two views
- 8-bit and 10-bit 4:2:0 color sampling
- Lookahead functionality
- Delta-QP support
- Separate bitrate allocation for secondary views
These features provide you with flexible options for encoding multi-view content while maintaining high quality and efficiency.
Technical implementation
The MV-HEVC implementation uses advanced techniques to achieve superior compression:
- Inter-view prediction: Reuses data from different views for maximum compression.
- Coding tools: Employs motion-compensated prediction and disparity estimation.
- Scalability: Implements a base layer for backward compatibility and enhancement layers for improved quality.
- Bitrate allocation: Implements separate allocation for secondary views, allowing application developers to experiment and fine-tune quality.
For more information about implementation, see the NVIDIA Video Codec SDK documentation.
Quality gains
The addition of MV-HEVC support has resulted in significant quality improvements, with bitrate savings of up to 27% as compared to simulcast encoding of two views.
Figure 3 shows the VMAF bit rate savings for the same clip.
Real-world applications
MV-HEVC is particularly useful for the following applications:
- 3D movies: Enhancing the viewing experience with efficient bit distribution between left– and right-eye views.
- AR/VR: Creating immersive environments with reduced data requirements.
- Spatial video: Supporting new technologies for capturing and viewing immersive 3D videos using advanced smartphones and specialized headsets.
- Auto: Process and compress multi-camera views for tasks such as objection detection and tracking.
The \Samples\AppEncode\AppEncCuda sample application in the Video Codec SDK now supports MV-HEVC encoding, providing you with a powerful tool for implementing multi-view video compression. For more information about implementation details for MV-HEVC support in AppEncCuda, see the NVIDIA Video Codec SDK documentation.
Backward compatibility
The NVIDIA implementation ensures backward compatibility, supporting two views with identical GOP structures for efficient random access and seeking. This approach lays the foundation for future expansion, as it can be readily adapted to accommodate more views in more complex multi-view applications. By maintaining consistent GOP structures across both views, we optimize synchronization and facilitate seamless navigation within the video content.
Conclusion
The addition of MV-HEVC support in NVIDIA VIDEO CODEC SDK 13.0 marks a significant milestone in multi-view video coding capabilities. With impressive BD-Rate gains and no performance overhead, this update provides you with a powerful tool for creating high-quality stereoscopic and 3D video experiences.
We encourage you to explore this new feature and integrate MV-HEVC encoding into your applications to take advantage of the improved compression efficiency and quality.
