Seedance image to video: animate an approved frame
Image-to-video works best when the still image already resolves the subject, composition, lighting, and visual identity. The prompt can then focus on movement instead of asking the model to redesign the frame and animate it at the same time.
Lock the visual decisions in the image, describe only the required motion, and review identity and geometry before export.
Prepare a stable start frame
Use a permitted PNG, JPG, or WebP image that is sharp enough to inspect at the intended output size. The current Try Seedance upload control accepts a start image up to 30 MB and can show an optional end-frame slot when the selected workflow supports it.
Remove accidental text, cropped limbs, distorted products, duplicate objects, or unresolved background geometry before animation. Motion generation can amplify problems already present in the source frame.
- Keep the subject fully visible with useful margin for movement.
- Choose the intended landscape, portrait, square, or wide composition first.
- Use only material you own or are permitted to upload and animate.
Write motion, not a second image prompt
Describe the subject action, environmental movement, and one camera move. State what should remain stable. For example: preserve the bottle shape and label placement; add a slow rightward camera slide while soft steam moves behind the unchanged product.
If an end frame is available, use it to define the destination rather than packing several transitions into the text. Avoid competing directions such as a locked camera and a dramatic orbit in the same shot.
- Subject: one observable action.
- Camera: one move, speed, and direction.
- Environment: restrained secondary motion.
- Constraints: details that must remain unchanged.
Verify settings and review the result
Choose only the duration, ratio, resolution, and sound options shown for the selected model. These values are loaded from the live catalogue and can differ by provider or mode. The credit estimate shown before submission is the relevant cost for the task.
After generation, inspect the opening frame, subject identity, hands or product geometry, camera path, background consistency, text, logos, and final edit point. Save a successful prompt with its exact settings rather than assuming the same wording behaves identically across model versions.
- Compare the generated opening frame with the source image.
- Watch once for subject motion and again for background artifacts.
- Check rights, consent, disclosure, and provider terms before publishing.
Official sources to verify
External product details can change. Use the linked vendor source for the current model ID, parameters, region, price, and policy.
- ByteDance Seed model directory
Use the official model directory to verify how the current Seedance family is described and which releases are listed.
Frequently asked questions
Do I need an end frame for Seedance image to video?
No. A start frame is required in the Try Seedance image mode; an end frame is optional when the selected workflow exposes that slot.
Which image formats can I upload?
The current workspace accepts PNG, JPG, and WebP start images up to the limit shown beside the upload control.
Which duration and resolution should I choose?
Use only the options shown for the currently selected model. Availability and credit cost can change with the model, mode, duration, and resolution.
Why does the animated subject drift?
Simplify the action and camera instructions, strengthen the preservation constraint, and start from a cleaner frame with fewer ambiguous or partially hidden details.
Is Seedance AI an image generator?
Seedance is presented as an audio-video generation family, not a standalone still-image generator. Prepare or create the source image with an image tool, then use an enabled Seedance image-to-video workflow to animate it.