Skip to content

Pusa

Video diffusion with frame-level noise control

Visit official site

What Pusa is

Pusa is described as a video diffusion model that introduces frame-level noise control and handles text-to-video and image-to-video tasks as an open source solution.

What you can do with it

  • Generate a video from a text prompt
  • Convert a still image into a moving video
  • Interpolate frames for smoother playback
  • Create seamless loops and video transitions

Application scenarios

  • Generate a clip from a prompt such as a person playing basketball
  • Turn a supplied image into a social media video
  • Produce a short commercial with loops and transitions

Getting started

  • Clone the code repository with Git and install dependencies
  • Download model weights from Hugging Face or other sources
  • Run a text-to-video command with the model path and a prompt
  • Try different condition positions to adjust the result

Pros & cons

✓ What we like

  • An open source solution for video generation
  • Seamless loops, interpolation and transitions are listed together

! What to watch out for

  • Generation starts from text prompts or supplied images
  • Documented setup includes cloning the code, installing dependencies and downloading weights

FAQ

Which generation tasks are listed?

The material names text-to-video and image-to-video, plus interpolation, seamless loops, transitions and longer videos.

How is the model installed?

The documented steps are cloning the code, installing dependencies, downloading weights and running a text-to-video command.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review