Skip to content

FMA-Net

Joint video super-resolution and deblurring with one model

Visit official site

What FMA-Net is

The model is described as a deep learning model for video super-resolution and deblurring, restoring low-resolution and blurry video into high-resolution and sharp video. Flow-guided dynamic filtering and multi-attention iterative feature refinement let it handle large motion in video.

What you can do with it

  • Restore video through super-resolution
  • Deblur blurry video
  • Handle large motion in video
  • Apply a simple model structure

Application scenarios

  • Raise the resolution of low-resolution video so it looks clear
  • Deblur motion-blurred video for clearer frames
  • Raise the resolution of surveillance video to identify details

Who it is for

  • Video editing work
  • Video enhancement work
  • Video analysis work

Pros & cons

✓ What we like

  • Restores resolution and removes blur in one model
  • Handles large motion in video

! What to watch out for

  • Restoration targets low-resolution and blurry video as input
  • Large-motion handling relies on flow-guided dynamic filtering

FAQ

What does the model restore?

The material states low-resolution and blurry video are restored into high-resolution and sharp video.

How are large motions handled?

Flow-guided dynamic filtering and multi-attention iterative feature refinement are described.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review