This AI Fixes Blurry Video Without Slowing Your Streaming

When video is squeezed to travel over the internet, quality suffers. You've seen it in video calls or streaming: blurry faces, blocky patches, weird smearing. That's not your connection's fault entirely — it's the price of compression. The newer H.266 compression standard shrinks files better, but it still leaves those ugly artifacts behind.

Researchers from Vietnam created MDFI, an AI that fixes those imperfections after the video is decoded. It doesn't just smooth things over. It studies the video's moving parts, different frequency details, and the compression's own "prediction" data — info the encoder already used — to rebuild a cleaner, sharper picture. Think of it like a photo editor that looks at the original scene and the blurry photo, then rebuilds the missing detail.

In tests, MDFI beat other leading video-enhancement methods on both technical scores and visual quality. The team also built a public dataset of uncompressed videos, compressed versions at several quality levels, and the prediction frames, so other researchers can compare and improve. The code is open source, meaning developers can start using it now.

For you, this could mean streaming services deliver better picture quality at lower bandwidth, video calls look less like a bad JPEG, and downloaded movies take up less space without looking worse. This isn't a product yet, but it's a clear step toward a future where "streaming quality" and "smart usage of data" finally go together.