Skip to content

VideoGrain

Zero shot video editing at multiple granularity levels

Visit official site

What VideoGrain is

The technique is a diffusion based video editing method that tunes spatial temporal attention for multi granularity control. It addresses semantic alignment and feature coupling.

What you can do with it

  • Edit at category, instance or part level
  • Control edits through text to region guidance
  • Separate features by adjusting attention

Getting started

  • Download the open source code and data from the project page
  • Prepare a video and its text prompt
  • Choose a category, instance or part level

Pros & cons

✓ What we like

  • Zero shot editing needs no extra training data
  • Combining with SAM-Track is described

! What to watch out for

  • Editing needs a video and a text prompt
  • A granularity level is chosen before editing

FAQ

What levels are supported?

Category level, instance level and part level editing.

Does it need training data?

The material describes zero shot editing with no extra training data.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review