Wan multimodal video

Wan AI Video Generator

Choose Wan 3.0 or Wan 2.7, then add the inputs and settings for that version.

Input

Model version
Generation mode

Switching modes keeps your inputs. Only inputs in the current mode are submitted.

0 / 20000

For editing, state clearly what should change and what should remain unchanged.

Reference inputs are optional. You can generate a video without them.

Image inputs

Add up to 10 image references. Refer to them as Image1, Image2, etc. in your prompt.

Video inputs

Up to 5 MP4 or MOV clips · 100 MB each · 1–15 seconds each, 15 seconds total · input + output up to 30 seconds

Reference audio

Up to 5 MP3 or WAV clips · 15 MB each · 1–15 seconds each, 15 seconds total

Generate audio
Aspect ratio
Video duration
Resolution

Output

Your video will appear here

Choose a model version, add your prompt and optional references, then generate a video.

The model suite, explained

What is Wan?

Choose Wan 3.0 or Wan 2.7, then add the inputs and settings for that version.

Wan 3.0 supports 480P, 720P and 1080P output from 2 to 30 seconds. Wan 2.7 retains its original mode-specific settings.

Each request runs as an asynchronous generation task. After completion, FramePack stores the output privately so you can preview and download it from the same workspace.

Wan 3.0 / Wan 2.7

Create with the Wan video toolkit

Move from an original idea to a guided animation or targeted edit without changing tools.

Prompt-led direction

Describe subjects, action, camera movement, atmosphere and sound in a single production brief.

Flexible visual inputs

Start from a first frame, add an optional final frame, guide consistency with references, or transform a source video.

HD output controls

Wan 3.0 supports 480P, 720P and 1080P output from 2 to 30 seconds. Wan 2.7 retains its original mode-specific settings.

Tracked private results

Follow the provider task while it runs, then preview and download the privately stored result.

Three focused steps

How to generate a Wan video

Choose a version, describe your video and add references as needed.

  1. 01

    Choose a model version

    Choose Wan 3.0 or Wan 2.7, then add the inputs and settings for that version.

  2. 02

    Add direction and references

    Write the prompt, upload the required media, then choose resolution, aspect ratio and duration.

  3. 03

    Generate, preview and download

    Submit the task, follow its status, and open or download the finished video when processing completes.

Wan prompting tips

Clear prompts and purposeful references make motion and edits easier to control.

  • Describe the subject, action, camera path, lighting and timing in the order they should appear.
  • Use clean, consistent images and explain which identity, object, style or framing each reference should preserve.
  • Choose the aspect ratio for the final channel first, then compose the prompt for that frame.
  • For video edits, separate what must stay unchanged from the exact transformation you want.

Questions, answered

Wan FAQ

What can Wan do on this page?

It supports text-to-video, first-frame and optional last-frame image-to-video, reference-to-video, and prompt-based editing of an uploaded video.

Which reference files can I upload?

Wan 3.0: JPG, PNG, WebP or BMP · 20 MB each · sides 240–8,000 px · aspect ratio up to 8:1; Up to 5 MP4 or MOV clips · 100 MB each · 1–15 seconds each, 15 seconds total · input + output up to 30 seconds; Up to 5 MP3 or WAV clips · 15 MB each · 1–15 seconds each, 15 seconds total

How long does generation take?

Timing varies with resolution, duration and provider queue capacity. The page keeps checking the asynchronous generation task while it runs.

Where is the finished video stored?

After the generation service returns the result, FramePack copies it into private media storage and provides authenticated preview and download links.

Turn an idea or reference into motion

Choose a Wan workflow, add clear visual direction, and create your next video online.

Create with Wan