ByteDance Seedance / VIDEO MODEL
Reference-led scenes, with controls for framing, motion, and sound.
Generated videos are stored for 24 hours, then deleted. Download files you want to keep.
Your next take will appear here.Sample footage · Seedance 2.0 workspace
No generated videos yet.
Your inputs are saved while you work.MEET SEEDANCE 2.0
Seedance 2.0 brings text to video, image to video and multiple reference inputs into the EveryGen AI workspace. Start from a written idea when the composition is open, an image when the first shot is already defined, or a reference collection when different files describe different parts of the scene. Your input mode determines which upload controls appear.
The useful decision is what each source should contribute. A portrait can describe a subject, a location image can establish the setting, and a short clip can suggest movement. Keep those roles in your prompt so you can judge the result against a specific brief. The current configuration offers four output sizes, a generated-audio option and clips from four to fifteen seconds.
THE POSSIBILITIES
Choose the input that fits your idea, then shape the motion. Clips below are local sample footage for the layout.
Describe the subject, its action and the camera path in that order. A single scene with one clear movement is easier to revise than a request containing several unrelated locations. Seedance 2.0 text to video needs a written prompt; images are optional only after switching to an image or reference mode.
Try a slow approach, a lateral tracking move or a stationary composition before adding more actions.
Use the opening image to establish the composition. An optional ending image describes the destination of the shot. For a transition between two frames, keep the subject, scale and setting compatible, then explain what changes between them. This is a different job from supplying several pictures as identity or style references.
Choose Reference images when the files describe appearance rather than two endpoints.
The reference workspace exposes image, video and audio uploads supported by this route. Explain which reference defines the subject and which defines movement; remove files that give conflicting instructions. When generated audio is enabled, add a short description of ambience or effects tied to the visible action rather than an unrelated soundtrack request.
Reference clips affect the quoted duration. Review the estimate after adding or replacing one.
FROM IDEA TO A TAKE
A clear starting point, a focused brief and settings you can review before submitting.
Open Seedance 2.0 in the generator above. Choose Text to video for a written scene or Image to video when you have an opening composition. Choose Multi-reference when several files have distinct roles.
Write the subject, action, camera direction and lighting. In image mode, add a first frame and an optional compatible last frame. If using appearance references, select the reference-image control and assign roles to up to 9 images.
Choose 480p or 720p or 1080p or 4K output. Pick a supported duration from 4 to 15 seconds. Set the aspect ratio for the destination. Review the estimate after changing a setting or adding a reference clip.
When generation is enabled, submit after reviewing the brief and content checks. Track the request in Recent takes. Review the full clip for subject details, continuity and sound, then download a useful result or change one part of the brief for the next take.
A STARTING POINT, NOT A SCRIPT
Load a brief with supported settings in the full workspace. Review it and add your own images or references where required.
Two friends walk along a seaside path at sunset. Track beside them at walking pace, keeping both faces visible. Warm natural light, gentle wind moving their jackets, a continuous shot with a steady horizon. Ocean ambience, no captions or scene cuts.
Keep the action and camera direction separate so you can change the pacing without rewriting the subject.
Use the first image for the person's identity and clothing, and the second image for the interior lighting. Use the reference clip only to guide the camera's movement. The person turns toward the window, then holds the pose. Keep the room layout stable and do not introduce another person.
Upload the mentioned references before estimating. File order should match the roles described in your prompt.
PUT THE SHOT TO WORK
Start with the job the clip needs to do. Keep final editing and a full review in your workflow.
Bring a clear identity reference and describe one gesture or camera move. Review faces, clothing and background details across the whole clip before keeping a take.
Separate the product reference from the environment and movement references. Keep exact labels and final advertising copy for a later editing step.
Choose compatible first and last frames for a reveal or change of lighting. Explain the intermediate motion so the ending frame has a reason to appear.
BEFORE YOUR NEXT TAKE
If two images show different clothes or lighting, decide which one should take precedence. State that decision explicitly. A smaller set with clear roles is more useful than uploading every available file without explaining why it is there.
Keep the inputs and framing the same while changing one camera instruction. Save the useful settings in your history. A seed can help document the request, but it does not promise an identical result or perfect character preservation.
THE SETUP
Every model has its own inputs and credit rules. Changing a setting refreshes your quote.
| Duration | 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 seconds |
|---|---|
| Output sizes | 480p · 720p · 1080p · 4K |
| Frame inputs | First frame + optional last frame |
| Additional references | 9 images · 3 videos · 3 audio files, where the mode supports them |
| Audio | Optional generated audio |
FIXED CREDIT RULES
A 20-credit content-check allowance is included once per request. Reference videos use the total uploaded duration, rounded up to a second.
| Resolution | Input | Rate |
|---|---|---|
| 480p | Text / supported images | 68 credits / output second |
| 720p | Text / supported images | 146 credits / output second |
| 1080p | Text / supported images | 363 credits / output second |
| 4K | Text / supported images | 739 credits / output second |
| 480p | With reference video | 41 credits / second of reference + output |
| 720p | With reference video | 88 credits / second of reference + output |
| 1080p | With reference video | 221 credits / second of reference + output |
| 4K | With reference video | 455 credits / second of reference + output |
Rate version v8-fixed-2026-10-11. The final quote appears before submission. Paid generation is awaiting activation.
Compare plans and credit allowances →BEFORE YOU CREATE
Inputs, output settings and credit rules for this workspace.
Text to video is supported: describe a scene and its movement. First frame + optional last frame are supported in image mode. A separate reference-image control accepts up to 9 images for appearance guidance.
480p, 720p, 1080p, 4K output is offered with durations from 4 to 15 seconds. The controls above use these same settings. A resolution selection describes the requested output and does not guarantee native detail or readable small text.
Multi-reference mode exposes up to 9 image, 3 video and 3 audio references. Explain each file's role in the prompt. Video-reference duration changes the estimate; inspect the quote after uploads are processed.
The credit table below the controls section lists this model's fixed rates for every supported output size. Some routes charge per request and others by duration. A content-check allowance is included once. Use Review credit estimate in the generator for the actual settings and uploaded reference duration before submitting.
This route has an optional generated-audio control under Audio, seed & camera. Describe relevant ambience or effects in the brief, then listen to the full result. Sound generation does not replace a final voice recording or music mix.
Prompts and image references are checked before submission. A rejection or unavailable check stops submission. These checks do not audit reference-video frames or audio tracks. Confirmed failed tasks release reserved credits once; an uncertain submission remains under review to avoid duplicate requests.
The current clips are local sample footage used to demonstrate the layout. They are not outputs from the displayed prompts, model benchmarks or evidence of a specific model's quality. Example prompts are editable starting points; image and reference examples require your own uploads.
FIND THE RIGHT FIT
Explore an idea and refine your next take with faster iteration.
480p / 720p · 3 input modesBring a focused scene and reference image into a video workflow.
720p / 1080p / 4K · 2 input modesYOUR NEXT SCENE
Start with a brief. Choose the inputs. Review the settings.
