Seedance guides
How to Use Seedance AI: From First Prompt to Download
Choose a video workflow, write a useful prompt, check settings and credits, and review your first Seedance clip.
Seedance is ByteDance’s video model family. sdanceai is an independent service that provides access to available Seedance models, alongside other creative tools. This guide takes you from an idea to a downloaded clip. You do not need an API key to use the web generator.
Choose the right starting point
- Text to Video: start with a written scene when you do not have a source image.
- Image to Video: animate a photo, product image or illustration you have permission to use.
- Reference to Video: use compatible image, video or audio references when you need more guidance over the scene.
Open the tool and choose an available Seedance model. Model versions and service providers can expose different options. The controls shown for your selected model are the current reference for duration, resolution, aspect ratio and audio. A specification on a model introduction page is not a promise that every workflow offers it.
Prepare one simple shot
For a first attempt, use one subject, one action and one camera direction. A five-second shot is a useful starting point if the selected model offers it. Avoid a whole advertisement with multiple locations in your first generation.
For text-to-video, try this starting prompt:
A ceramic coffee cup stands on a wooden kitchen table at sunrise. Steam rises slowly from the cup. The camera makes a gentle push-in in one continuous shot. Soft warm light comes from the window. Quiet room ambience, no dialogue.
For image-to-video, upload a clear product photo and describe motion rather than redesigning the product:
Use the uploaded bottle as the subject. Keep it upright and preserve its shape and packaging. The camera slowly moves closer. A soft reflection travels across the glass. The background stays still. No hands, no dialogue.
These are illustrative prompts to adapt, not measured performance examples. Read the prompt guide for more variations.
Pick output settings
Choose a publishing ratio first: for example, 9:16 for a vertical clip or 16:9 for widescreen, when offered. Select the supported duration and resolution. Start with a short clip to check the motion before spending credits on a longer version.
If the model exposes native audio, decide whether you need it. Describe desired ambience or dialogue explicitly; sound instructions alone do not enable an audio option that is turned off or unsupported. Reference files must meet the upload area’s current type, size, count and duration limits.
Sign in and check the credit estimate
Sign in before submitting. Check your balance and the estimate for the exact model and settings. Do not use a model’s lowest unit-price label as the total cost of your video. Reference media and optional controls may change the quote.
The pricing page can estimate a batch of text-only clips. Subscriptions and one-time credit packs are separate purchase choices; a calculated share of a plan is not an individual video purchase price. Trial access depends on your current account offer. See the trial guide before expecting a free generation.
Generate, inspect and download
Submit once, then follow the task status. A queued or processing task is still running. Repeatedly submitting it can create separate billable jobs. When it completes, review the clip in the tool’s work/history area and use the available download action.
Check these details before sharing:
- Does the subject perform the intended action?
- Do the face, product shape, packaging and text remain acceptable?
- Is the camera movement useful, and does the framing fit your platform?
- If audio is enabled, are the sound and any spoken words suitable?
For another attempt, change one variable. Make a weak action more explicit, reduce complex movement if a face changes, or remove conflicting camera directions. A completed render that you dislike is different from a failed task. Failed tasks follow the credit-return policy; an unattractive completed result is not automatically a failed generation.
Make a longer story and publish responsibly
If your story exceeds the available single-clip duration, write separate shots, reuse your references and assemble the clips in your video editor. Inspect transitions instead of assuming independent clips will join perfectly.
Commercial use follows the Terms of Service, payment of required credits, applicable model terms and the rights to your inputs. Subscription credits and one-time credit packs follow the same usage terms. Usage permission does not guarantee exclusive copyright or clear third-party rights.
For account or task problems, contact support@sdanceai.com with the task ID and selected settings. Never send your password or API key.