How to Use Wan 3.0

Learn how to use Wan 3.0 in CawCut for multimodal video generation, reference-guided scenes, and prompt-led video editing.

Updated Nov 18, 2026 · 7 min read

Build a video from a written idea, visual references, or a combination of inputs—and use Wan 3.0’s Thinking option when you want an extra reasoning pass before generation. Wan 3.0 is available directly in CawCut’s [Video Generation] node, with support for text, images, frames, and omni references.

1. Wan 3.0 Strengths

StrengthApplication
Flexible multimodal generationUse text, images, first/last frames, or omni references to guide a scene.
Video editing directionDescribe the change you want in an existing video for prompt-guided replacement or removal workflows where available.
Reasoning before generationEnable Thinking when you want the model to reason through a more involved prompt before producing the clip.
Reference-led controlCombine a clear prompt with visual guidance to shape subject, setting, and movement.

2. Modes and Inputs

Wan 3.0 supports Text, Image References, Frames, and Omni References in CawCut. Its Thinking option is available for Wan 3.0.

ModeUse it forTypical input
[Text]Creating a scene from a written brief.Prompt describing subject, action, setting, and camera.
[Image References]Animating or guiding an existing visual.Image or character reference with a motion direction.
[Frames]Controlling the opening and optional ending composition.First frame, optional last frame, and transition direction.
[Omni References]Coordinating several media types in one generation.Supported images, character, video, or audio references plus a prompt.

3. Prepare Prompts and References

Assign one job to each reference: establish a subject, set a first or last frame, guide motion, or supply broader visual context. Write actions in sequence and make the intended change explicit when editing a clip.

What to avoid: Avoid contradictory references, unclear instructions about which elements to keep, and prompts that combine multiple unrelated edits. Do not ask one image to serve as both a strict composition lock and a completely different scene design.

For a clear prompt:

  1. Identify the main subject and the key action.
  2. Put events in the order they should happen.
  3. Give the camera one coherent direction.
  4. State which subject or scene details should remain consistent.
  5. Add style and audio direction only when they support the intended result.

4. CawCut Workflow and Settings

  1. Add a Video Generation node and select Text, Image References, Frames, or Omni References.
  2. Choose Wan 3.0 and connect the references that support the selected mode.
  3. Describe the scene or edit in order; identify what should change and what should remain stable.
  4. Set the available duration, aspect ratio, resolution, and audio options. Turn on Enable Thinking when the prompt benefits from a reasoning pass.
  5. Review the credit estimate, generate, and inspect the result for visual continuity, motion, and audio.
  6. If refining, adjust one reference or one instruction at a time.

For full mode and setting details, see Video Generation. For connecting this node with other steps, see Understanding Nodes.

5. Limitations and Refinement

Complex instructions can still be interpreted differently from your intent, especially when references conflict or a prompt combines several changes. Review edits carefully and use a focused follow-up direction for missed details.

If you noticeTry this
The action is hard to followBreak it into ordered beats and focus on one key movement.
The subject or visual style driftsUse a clearer reference and list the details to preserve.
Camera movement is confusingSimplify it to one camera direction and one main action.
A detail is almost rightKeep the successful inputs and refine only that element.

Review the full clip before using it as final footage. Check Credits before running many variations.

Related articles

Frequently Asked Questions

Learn how to use Wan 3.0 in CawCut for multimodal video generation, reference-guided scenes, and prompt-led video editing.

It is a flexible video model for text, image, frame, and omni-reference workflows, including prompt-directed changes to video.

Text, Image References, Frames, and Omni References.

It enables Wan 3.0’s thinking mode before video generation. Use it for prompts that benefit from a reasoning step.

Wan 3.0 is recommended in CawCut for prompt-guided removal or replacement of elements in an existing video. Use the mode and inputs shown in the node.

Explain the role of each reference and avoid giving different inputs contradictory instructions.

Name the element to change, describe the desired result, and list what must remain unchanged. Refine one edit at a time.