Turn text, images, video and audio references into cinematic AI videos online with Seedance 3.0.
Seedance 3.0 turns text descriptions and reference files into video that unfolds the way you intend: what the subject does, how the camera moves and when the pace changes can all be specified in your prompt and references. Everything you create in the Seedance 3.0 workspace is kept together on the same page.
A single prompt can carry an opening shot, a change of camera angle, the main action and a closing beat, which suits product reveals, short story moments and social clips that need a beginning and an end rather than a looping fragment. Choose a fixed length when timing matters, or let the model decide how long the scene should be.
Choose the resolution that fits the job: lower resolutions suit quick tests and side-by-side comparisons, while higher resolutions suit final publishing, large screens and cropping in post. The workspace shows the cost before you submit.
Combine images, video clips and audio in one task. Images can fix a character, a product or a visual style. Video clips can show a camera move or a motion pattern. Audio can set rhythm, mood or spoken lines. In your prompt, refer to each file by its position, such as @1 or @2, so the model knows which reference controls which part of the scene.
Beyond generating from scratch, you can upload an existing video as a reference and continue or refine it. Decide what the task should achieve before you write the prompt, and the result will be closer to what you expect.
Not in full yet. The current Seedance 3.0 is a testing version built on Seedance 2.5, focused on optimizing a number of specific scenarios, and is not the full version. The full version is expected to launch in February 2027. Nothing you upload or generate during testing is used to train AI models.
The full version is expected to open in February 2027. This page will be updated as soon as it is available.
For now, testing access is open only to subscribers on the annual Pro and annual Max plans. Accounts without testing access can browse the prompt guide and examples on this page.
During testing, Seedance 3.0 is not part of the free allowance. Creating an account is free, and Seedance 2.0 Mini is free every day within its daily allowance, so you can start creating without a paid plan. The Pricing page lists what each plan includes.
No. We do not use your uploaded images, videos or generated content to train AI models. Your content is used solely to provide the generation you request, as set out in our Privacy Policy.
Privacy PolicyThe Seedance pages on this site suit different kinds of work. Choose according to what the scene needs.
Designed for longer scenes with several shots, directed through a combination of image, video and audio references. Choose it when the scene needs control over characters, camera and pacing.
Suits shorter single clips created from text, images or audio, and everyday projects that do not need a multi-shot structure.
The quickest and lowest-cost option, with a free daily allowance. Use it to test an idea before spending credits on a larger scene.
The workspace follows a real production workflow. Start with a single description, then add character images, a camera reference or background music, and refine until the result matches your idea. The sections below cover the creative features, the steps to generate, prompt writing, camera language and reference preparation, so you can reach stable, controllable results faster.
Describe the subject, the action, the setting and the camera. Short, specific sentences work better than long lists of adjectives.
Upload images, video or audio, then mention them in the prompt by position.
Generate a new clip from scratch, or build on an existing video.
The workspace shows the credits required before you submit, so you can adjust the settings first.
Track progress in the workspace. Finished videos are saved in My Videos on the Seedance 3.0 page, where you can download them or reuse their settings.
When writing a Seedance 3.0 prompt, you can usually organize it in a simple order: subject, action, environment, camera and style.
A ceramic coffee cup on a marble counter, steam rising slowly, morning light from the left, the camera pushes in from a medium shot to a close-up, warm and calm.
A prompt like this gives the model every decision it needs without contradictions.
For longer clips, describe the scene in beats. Start with what the viewer sees first, then state what changes, and end with the final frame. Time words such as "first", "then" and "finally" help the model distribute action across the whole clip instead of rushing everything into the opening moment.
When you use references, give each file a job. "@1 is the main character, @2 sets the lighting style, follow the camera movement in @3" is far more reliable than attaching three files and hoping the model infers their purpose. Avoid asking one reference to do two conflicting things, such as defining both the character and the background.
Keep negative instructions short. Listing many things to avoid often draws attention to them. Describe what you want to see instead.
The prompts below follow the same order of subject, action, environment, camera and style. Use them as starting points and replace the details with your own.
A matte black wireless headphone rests on a dark stone pedestal. A thin beam of light sweeps across its surface and reveals the texture of the ear cushion. The camera orbits slowly from the side to the front, then pushes in on the ear cup. Deep shadows, cool highlights, a premium and restrained look.
One subject, one change of light and one continuous camera move keep the product in focus from the first frame to the last.
A young woman in a yellow raincoat waits at an empty bus stop at dusk. First she checks her watch, then she looks up as headlights appear in the distance, and finally she smiles and steps forward. The camera begins on a wide shot and ends on a close-up of her face. Light rain, warm street lamps, a quiet and hopeful mood.
The words "first", "then" and "finally" divide the action into beats, so the scene has a beginning, a turn and an ending.
@1 is the main character and @2 sets the color palette. She walks through a night market, stops at a food stall and turns towards the camera. Follow her with a handheld tracking shot at shoulder height. Keep her clothing and hairstyle exactly as they appear in @1.
Each reference has a single job, and the prompt states which details must stay unchanged.
A mountain lake at sunrise, with mist drifting above the still water. The camera rises from the shoreline, glides over the lake and reveals snow-covered peaks in the distance. A small red canoe crosses the frame from left to right. Natural colors, clear air, a calm documentary style.
A single rising camera move reveals the scene gradually, and one small moving element gives the frame a sense of scale.
Naming the camera movement in your prompt helps the result come closer to what you have in mind. The descriptions below are the most common moves and when to use them. Use one on its own or combine several in sequence.
The camera moves toward the subject, for example from a medium shot to a close-up. Use it to emphasize product detail or a character's emotion.
The camera moves away from the subject and reveals the surroundings. It works well for opening a scene or closing it.
The camera circles the subject to show several angles, which suits product showcases and character reveals.
The camera turns left, right, up or down from a fixed position to sweep across a scene or guide the viewer's eye.
The camera follows a moving subject and keeps it in the same place in the frame, ideal for walking, running and driving scenes.
A slightly shaky camera creates an authentic, in-the-moment documentary feel that suits social media videos.
The camera rises from a low point or looks down from above, which suits cities, landscapes and large-scale scenes.
The camera stays still so the action in the frame becomes the focus, which suits interviews, still lifes and slower moments.
Good references matter more than many references. A few habits keep Seedance 3.0 results consistent from one scene to the next.
Use well-lit images in which the subject fills most of the frame, so the character, product or style is easy to read.
Trim video references to the motion you actually want, because unrelated movement can leak into the result.
Choose clean recordings without heavy background noise, so rhythm, mood and spoken lines come through clearly.
For a series with the same character or product, keep a small folder of approved references and reuse it, so every scene starts from the same visual identity.
Show a product from several angles, add a branded color palette through reference images and deliver a short spot without a studio shoot.
Create vertical clips with a clear hook in the first seconds, then develop the best ones into full versions.
Test camera moves and blocking for a live-action shoot before booking crew and locations.
Turn a short script into a visual sequence that holds attention longer than still slides.
Use an audio reference to guide pacing so the movement follows the rhythm.
Answers about Seedance 3.0 credits, mobile use and getting consistent results.
Seedance 3.0 is an AI video generator that turns text descriptions and reference files into multi-shot videos. You describe the subject, the action, the setting and the camera, add images, video clips or audio when you need more control, and direct how the scene develops from the opening shot to the closing frame.
Seedance 3.0 accepts a text prompt on its own, or a prompt combined with images, video clips and audio. Images usually define a character, a product or a visual style; video clips show a camera move or a motion pattern; audio sets rhythm, mood or spoken lines. In the prompt, refer to each file by its position, such as @1 or @2.
Yes. Content created on these plans includes a commercial license: all annual plans, monthly Standard / Pro / Max, all team plans, and credit packs from Creator Pack up. Content from the free allowance, monthly Basic, Starter Pack and trial packs is for personal, non-commercial use only. Also make sure the reference materials, brand elements, likenesses and audio you upload are properly licensed.
Yes. The workspace is designed for mobile browsers, so you can write prompts, upload images and videos, submit generations, and view or download your work in My Videos from your phone.
Failed tasks are refunded automatically. Your balance may take a moment to update, and you can confirm the refund on the Usage page.
Prepare a fixed set of reference images for the character, such as front, side and full-body shots, upload the same set every time, and state in the prompt that @1 is the main character. Describe the character with the same wording for clothing, hairstyle and colors instead of rephrasing each time. For a series, reuse the settings of the previous video from My Videos and change only the scene and action.
First check that the prompt covers the subject, action, setting and camera. If the description is too long or contains conflicting requests, simplify it and try again. When you use references, state what each file is for. For complex scenes, change one thing at a time so you can see what made the difference.
Browse curated Seedance 2.0 video prompt examples with videos, authors, publish dates, and engagement metrics.
