Uploaded October 2025 | Updated September 2026, 1 week ago
In this presentation, we will talk abut how Meta's Media Foundation team deployed advanced video super resolution (VSR) technology at scale to enhance video quality across Meta's platforms and multiple use cases.
We will explain Meta's comprehensive super resolution strategy, which applies AI-powered enhancement both server-side during video ingest and client-side during playback to improve resolution and visual quality.
We will than discuss two use cases where we deployed VSR for ads videos/images and Generative AI video workflows.
We will conclude with key learnings from leveraging large-scale subjective evaluation and objective metrics (UVQ) to measure quality improvements and detect artifacts, guiding model selection and deployment.
In this presentation, we will talk abut how Meta's Media Foundation team deployed advanced video super resolution (VSR) technology at scale to enhance video quality across Meta's platforms and multiple use cases.
We will explain Meta's comprehensive super resolution strategy, which applies AI-powered enhancement both server-side during video ingest and client-side during playback to improve resolution and visual quality.
We will than discuss two use cases where we deployed VSR for ads videos/images and Generative AI video workflows.
We will conclude with key learnings from leveraging large-scale subjective evaluation and objective metrics (UVQ) to measure quality improvements and detect artifacts, guiding model selection and deployment.
