Skip to content

Wan2.1

Open-source video generation across text, image and editing tasks

Visit official site

What Wan2.1 is

The model is described as an open-source large-scale video generation model built with a spatio-temporal VAE, scalable training strategy, large-scale data construction and automated evaluation metrics. The material states it supports text-to-video, image-to-video and video editing tasks, and notes strong results on several benchmarks.

What you can do with it

  • Generate video from a text description
  • Generate moving video from a still image
  • Edit and refine existing video
  • Generate video content with Chinese and English text
  • Encode and decode 1080P video with a video VAE that keeps temporal information

Example uses

  • Generate a video of two human-like cats boxing on a stage from a text description
  • Turn a still beach photo into a moving beach scene with waves, sunlight and sand
  • Raise a low-resolution video to high resolution while improving picture quality

Getting started

  • Clone the repository from the GitHub project
  • Install the dependencies listed in requirements.txt
  • Download model weights through Hugging Face or ModelScope
  • Run generate.py with the task type, model path and input parameters

Pros & cons

✓ What we like

  • Several video tasks are supported in one model
  • The model is open source and can be freely used and extended

! What to watch out for

  • Generation tasks rely on the model weights downloaded separately
  • Running the model follows a script with task type and input parameters

FAQ

Which tasks are listed?

The feature list names text-to-video, image-to-video, video editing and multilingual text generation.

What model weights are needed?

The documented setup downloads weights through Hugging Face or ModelScope before running the generation script.

Last reviewed: 2026-11-15

More AI video generation tools

View all →

How we review