Seedance 2.5 AI Video Generator

Seedance 2.5 creates a video from your prompt and optional image references. Choose one or more catalog models, upload your own human reference photos, or use only products, outfits, objects and scenes.

How to use Seedance 2.5

Optionally choose catalog models or upload your own human photos. Add other image references separately. Use up to eight references in total; JPG, PNG and WebP uploads must be under 10 MB each.
Describe your scene. Type @ to select @Image1, @Image2 and the other references shown in API order. Prompts may contain up to 30,000 characters.
Select 5–30 seconds, a supported aspect ratio and 480p, 720p or 1080p. Review credits, then generate.

Credits per second

Choose a character: 11 / 23 / 50 credits per second. Upload your photo: 23 / 46 / 100 credits per second. Genjutsu: 16 / 35 / 80 credits per second. Resolution order: 480p / 720p / 1080p. Total credits = rate × duration.

Frequently asked questions

How do I use Seedance 2.5 on Retwist?

Choose your character, write the scene prompt, set duration and output settings, then click Generate. The result is saved in Studio.

What is the difference between character selection and photo upload?

Catalog models and requests without uploaded human references use BytePlus. Uploading one or more human reference photos uses fal.ai. Credits follow the selected provider, duration and resolution.

How should I refer to my character in the prompt?

Use the @Image numbers displayed on the thumbnails. Human references come first, followed by other images. Type @ in the prompt to select a reference, and explain how each should be used.

What durations can I select?

Select any whole number from 5 to 30 seconds. This is the requested video duration, not the processing time.

Which aspect ratios are available?

Choose 9:16, 16:9, 1:1, 4:3, 3:4 or 21:9. Vertical 9:16 is the default. These are fixed output ratios shared by the reference generation setup.

Which resolutions and bitrate settings are used?

Choose 480p, 720p or 1080p; 480p is the default. fal.ai requests high bitrate automatically. BytePlus uses its documented output encoding; the fal.ai bitrate_mode control is not a BytePlus API parameter.

How much does a Seedance 2.5 video cost?

Credits equal duration multiplied by the current provider and resolution rate. Choosing a character and uploading your own photo have separate rates. The creator shows the current total; admin pricing changes can require you to review an updated price.

What photo should I upload?

Upload clear, authorized JPG, PNG or WebP references under 10 MB each. Add people through Human references and objects or scenes through Other images. Photos containing people in Other images may be rejected by the provider.

How long does generation take?

Processing time varies with provider load and the request. You can leave the page and check Studio; the selected duration describes the output clip.

What happens if the provider rejects my request?

Studio shows the failure or review status. A confirmed failure follows the refund rules. An uncertain submission is reviewed to avoid charging or submitting the same job twice.

Can I change a source video here?

This page generates a new scene from a character photo and prompt. Use Higgsfield Genjutsu for source video motion transfer and reference based replacements.

Do I need to provide a human reference?

No. Human references are optional. You can use only other images, or describe a video with a prompt alone. Without uploaded human references, the request uses BytePlus pricing.

How many models or photos can I use?

Use up to eight references in total, including catalog models, human photos and other images. Choose multiple catalog models or upload multiple human photos; these human source options are separate.

What happens if I remove a referenced image?

Later @Image numbers are updated to match the new order. Mentions of the removed reference are flagged; revise those prompt assignments before generating.

Retwist.ai