Open Sourve pVMAF
Synamedia
Category QC QA Monitoring
While VMAF (the popular video quality metric released by Netflix) is great for VOD environments, it is too computationally heavy for live streaming. Synamedia’s pVMAF (predictive VMAF) is a lightweight, in-loop video quality metric that delivers VMAF-level accuracy yet, unlike VMAF, with minimal computational overhead. This opens up VMAF quality measurement to anyone live streaming video. With the goal of encouraging community collaboration and further optimisation, Synamedia unveiled this open source version in December 2024.
pVMAF uses key encoding parameters and pre-analysis statistics to achieve VMAF-like precision, but without the computational load. This makes it a game-changer in real-time video quality measurement and compression operations. By simplifying workflows, lowering costs, and future-proofing systems, it sets a new standard for video encoding solutions.
Using ML networks embedded within encoders, Synamedia is making strides in reducing CPU usage and optimising bitrate for improved video quality. The integration of pVMAF into this workflow is a critical component, as it allows for smarter decision-making in real-time, ultimately leading to a more efficient and cost-effective streaming experience.
pVMAF’s versatility makes it ideal for a range of applications:
• In-depth video quality analysis: pVMAF’s frame-by-frame evaluations reveal precisely which segments are more challenging to encode to predict where visual quality degradation occurs. This helps troubleshoot specific encoding issues.
• Live encoding monitoring: pVMAF alerts when video quality falls below a defined threshold. Automating quality checks reduces the need for human oversight, improving operational efficiency.
• Proactive rate control: pVMAF informs the rate controller if constant-quality encoding is not on target. This feature is now in Synamedia’s QC-VBR bitrate control algorithm, ensuring target quality without sacrificing efficiency.
