AI BriefingVMAF v1: Good Enough Isn't Enough
Key point
Netflix has open-sourced VMAF v1, which overcomes the limitations of the existing VMAF and improves detection of compression artifacts and responsiveness to viewing environments.
Details
Netflix has open-sourced v1, a new version of VMAF (Video Multimethod Assessment Fusion), which has established itself as the standard for video encoding optimization. The existing v0 model tended to be less sensitive to compression artifacts than to resolution degradation when bitrates were low.
VMAF v1 introduces the following key improvements to address these limitations.
- Improved compression artifact sensitivity: In addition to the existing DLM (Detail Loss Metric), an AIM (Additive Impairments) component that better captures visual disturbances such as blocking artifacts was added and linearly combined.
- Viewing environment optimization: Improving on the fact that the existing model was designed based on a living room environment (1080p@3H), the model was enhanced to enable accurate quality prediction across various screen sizes and viewing distances, including mobile devices.
Through this update, Netflix has laid the groundwork for more precisely measuring and optimizing visual quality even in a changing media landscape that includes next-generation codecs like AV2, as well as live streaming and cloud gaming.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Interaction-Aware Video Object Removal (VOID)