Seedance 2.0: Video Prompts, Audio and Settings

Seedance is ByteDance's video-model family. This page documents Seedance 2.0 Text to Video on FluxoKit, including the current duration, aspect-ratio, audio and output settings below. It also explains when an image or reference workflow is a better starting point. It does not treat the family name as a single interchangeable model.

A useful short clip starts with a shot you can describe and evaluate. For a U.S.-English product video, define the visible action, framing and any spoken words before generating. Do not ask a model to invent customer endorsements, product specifications or offer terms.

Choose the right starting material

• Text to video: start here when you need the model to construct the scene from your brief.
• Image to video: use an approved starting image when the composition is already chosen. See the image-to-video guide.
• Reference workflow: select a model that accepts the media you have when identity or scene references matter. Check its own input requirements rather than assuming the text-to-video form accepts those files.

A reference is guidance, not a continuity guarantee. Review the face, clothing, object proportions and background in the beginning, middle and end of each result.

Direct one readable action

Write the shot in this order: subject and setting, action, camera behavior, then sound. Keep the action small enough to fit the chosen duration. Several location changes, conflicting camera directions and a long script make it difficult to identify what went wrong.

Illustrative prompt to adapt, not a reported generated clip:

A person places an unbranded insulated bottle into the side pocket of a daypack beside a trailhead. Medium close-up, steady camera, soft early-morning light. One continuous action. Natural fabric and zipper sounds, no dialogue, no music, no text overlay.

The review is concrete: does the bottle enter the pocket, do the hand and pack remain plausible, and does the shot stay continuous? If not, simplify the action before adding more camera movement. This is a practical review method, not a claim that the prompt has been tested or will always succeed.

Set duration, framing and sound deliberately

Use the supported options in the parameters section. Choose the aspect ratio for the destination before the run, leaving space for captions and platform controls. Duration describes the requested clip length, not how long generation will take.

When audio is enabled, listen to the entire result. Check any dialogue against the approved script, confirm names and numbers, and reject unwanted speech or misleading claims. Add and verify captions for published spoken content. A silent clip should still communicate its action visually.

Compare outputs without inventing a winner

The published clip lets you inspect its motion and image details. Its original prompt and settings are not published in this gallery; the backpack prompt above is illustrative, not the source of that clip. The example does not prove that Seedance is the fastest model or that it beats Kling or Veo on every task. For your own comparison, use the same shot brief, crop and review criteria, and keep selected settings with each result.

Check credits before the next run

Review the studio estimate after setting duration, resolution, audio and any other options. The displayed estimate, reserved credits and final settled charge have different roles; generation history records the outcome. Do not infer a fixed cents-per-clip price from a subscription allocation. Current plans explain billing and included credits.

For the wider family, see ByteDance's model catalog. Provider releases and consumer-app features are not automatically available in this integration. FluxoKit's visible model selection and parameter table define the workflow described here; no completion-time, ranking or conversion result is promised.