Nokia releases public NNPF software to boost VVC video quality
Nokia has released public Neural Network Post Filtering (NNPF) software to improve VVC video quality through AI-based post-processing. The tool enables streaming platforms to restore detail and texture during final decoding using standardized metadata and inference on consumer hardware.
Key Takeaways
- Software applies AI-based post-filtering at the final decoding stage to improve perceptual quality, specifically targeting fine textures and gradients.
- Integrated support for VSEI messages and OpenVINO allows for real-time inference on a wide range of consumer-class hardware.
- Initial testing shows significant gains, including a reported 10.75% improvement in VMAF scores for specific 4K sequences like 'DucksTakeOff'.
- Release is governed by the BSD 3-Clause Clear License, lowering the entry barrier for encoder vendors and platform developers to experiment with VVC.
Why It Matters
Nokia’s move signals a shift from raw compression to AI-assisted reconstruction as the primary lever for video quality improvements. By aligning NNPF with existing VVC and VSEI standards, Nokia provides a non-proprietary framework that circumvents the fragmentation typically seen in AI video enhancement. For the ecosystem, this decouples visual fidelity from distribution bitrates, allowing streamers to offer 'premium' visual tiers without increasing egress costs. It also places the processing burden on the maturing AI acceleration blocks in modern consumer silicon. Watch for a rise in VVC-compliant test streams from major CDNs as they evaluate the impact on low-bandwidth subscriber retention.
Additional Context
The launch of Nokia’s NNPF software arrives as the industry reaches a critical juncture for Version 4 of the Versatile Supplemental Enhancement Information (VSEI) standard. Per Nokia reporting from November 2025, VSEI version 4 was specifically enhanced to handle AI and machine vision use cases, providing the synchronized metadata necessary for NNPF to operate across diverse device types without breaking backward compatibility. This standardization effort is part of a broader push to make VVC (H.266) the dominant successor to HEVC, particularly after the codec powered 8K streaming trials during high-profile international sporting events earlier in the decade.
Market adoption of VVC has historically been tempered by patent complexities and the rise of the royalty-free AV1 codec. However, recent developments have stabilized the licensing landscape. Per Streaming Media in March 2026, the launch of the Video Distribution Patent Pool by Access Advance in early 2025 established clear royalty rates for VVC, AV1, and HEVC, providing the fiscal predictability required for large-scale enterprise deployment. This stabilization has allowed technical committees, such as the Media Coding Industry Forum (MC-IF), to shift focus toward 'real-world' adoption through technical guidelines released in June 2026.
Hardware readiness has also converged with Nokia’s software release. In early 2026, industry analysts noted an 'inference flip' where global compute spend on running AI models officially surpassed training costs. This trend is visible in the streaming hardware sector, where latest-generation chips from NVIDIA and specialized mobile VPU vendors now feature dedicated neural processing units (NPUs). Per Netint’s 2026 infrastructure report, nearly 60% of video professionals have already integrated AI into their encoding workflows, with VVC adoption expected to ramp up as on-device AI acceleration becomes a standard feature in consumer smart TVs and mobile handsets.
Read full article at nokia.com
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source