Wan 3.0 multimodal reference control
Guide Wan 3.0 generation with up to 10 images, 5 video clips, 5 audio clips, or one document or public web page.
Understands your media, documents, and web pages, giving video generation richer creative context.
Wan 3.0 AI video generator
Wan 3.0 is an AI video generation model from Alibaba that turns written direction and reference materials into moving scenes. On Seedance 2.0, Wan 3.0 is available in a browser workspace with Text, Frames and References modes.
Guide Wan 3.0 generation with up to 10 images, 5 video clips, 5 audio clips, or one document or public web page.
Choose Wan 3.0 output at 480P, 720P or 1080P, with Adaptive, 16:9, 4:3, 1:1, 3:4 or 9:16 aspect ratios.
Write Wan 3.0 prompts up to 20,000 characters, use start and end frames, enable synchronized audio, and reuse a seed for repeatable settings.
Combine product images in Wan 3.0 and documents to turn features, positioning, and visual direction into a concise promotional video.
Use Wan 3.0 audio references to guide rhythm and mood, with images or video defining performers, scenes, and motion.
Direct Wan 3.0 characters, locations, camera language, and story beats from multiple visual and text references.
Turn a supported file or public web page into a product explainer, content summary, or presentation video.
Define characters, products, scenes, and visual style.
Guide motion, camera movement, and pacing.
Provide music, voice, rhythm, and atmosphere.
Add product facts, scripts, and story background.
Supply public page content and its surrounding context.
Upload Wan 3.0 reference media or a document, add a public web page, or choose start and end frames.
Write the desired result and use Image1, Video1, or Audio1 to point to specific media.
Select generation mode, duration, resolution, aspect ratio, and audio, then start generation.
Start a Wan 3.0 text to video project by identifying the subject and its main action. Specify the setting, framing, lighting and movement in that order, then describe the intended ending. For Wan 3.0, a single deliberate camera move is easier to evaluate than several competing movements. Reserve longer durations for actions that require visible progression. Review whether the subject completes the requested action before adding more locations or secondary characters.
Use Frames mode for Wan 3.0 image to video when the initial composition is already approved. Upload the required first frame and optionally supply a last frame to describe the intended destination. In the prompt, explain the movement between these positions: a camera approaching a display, a person turning toward a doorway or a product moving into light. Wan 3.0 still needs a plausible transition; distant viewpoints and unrelated subjects can make the motion difficult to direct. Select References mode instead when multiple images should guide identity or styling without defining exact endpoints.
A Wan 3.0 reference to video brief should explain why each uploaded asset is present. Use Image1, Video1 and Audio1 labels to connect instructions to the corresponding uploads. Wan 3.0 can receive several references, but adding more assets does not automatically produce a clearer brief. Keep the most important identity details consistent across references. This preparation makes it easier to judge whether a generated sequence follows the intended subject, motion and visual treatment.
| Model | Wan 3.0 | Wan 2.7 | Seedance 2.5 |
|---|---|---|---|
| Input parameters | Text · Images · Videos · Audio · Documents / Web link | Text · Images · Videos · Audio | Text · Images · Videos · Audio |
| Output format | MP4 | MP4 | MP4 |
| Duration (seconds) | 2–30 / Auto | 2–15; References: 2–10 | 4–30 / Auto |
| Native resolution | 480P / 720P / 1080P | 720P / 1080P | 480P / 720P / 1080P |
| Aspect ratio | Adaptive / 16:9 / 4:3 / 1:1 / 3:4 / 9:16 | 16:9 / 9:16 / 1:1 / 4:3 / 3:4 | 1:1 / 21:9 / 4:3 / 3:4 / 16:9 / 9:16 / Adaptive |
| Prompt limit (characters) | 20000 | 5000 | 10000 |
These original Wan 3.0 briefs illustrate a controllable subject, a defined movement and a clear ending. Adapt the references and settings to the actual project; the examples describe intended results rather than guaranteed output.
Use Image1 as the ceramic lamp reference and Image2 as the room reference. Preserve the rounded shade and matte ivory finish. Begin with a medium view of the lamp on a walnut desk. The camera moves slowly closer while warm light appears beneath the shade. Keep the surrounding objects still. Finish with the entire lamp visible and clear space on the right for a title. Include quiet room ambience without speech.
Use Image1 for the traveler and Image2 for the station. The traveler holds a blue notebook at waist height and walks toward the timetable. Use a steady side tracking shot with soft morning light. Preserve the coat color and notebook throughout the movement. End as the traveler stops and looks upward. Keep the background activity subtle. Include distant station ambience; no dialogue or additional music.
Use the uploaded product brief as factual context. Create a short visual introduction centered on the portable reading light described in the document. Show a desk at dusk, the light being positioned above an open book and a final view of the illuminated pages. Maintain a restrained editorial style with a stationary camera. Do not invent numerical performance claims. Leave detailed specifications and captions for the final edit.