How to use Seedance 2.0: a beginner's guide
Seedance 2.0 can be a starting point for creating a short scene from text or an image. In SWRFlow, the 2.0 family includes several model options, so first check the full model name. This guide explains the shared workflow and the choices to review before each request.
The starting point
Choose the exact Seedance 2.0 variant, select text or image input, describe one action and check the available settings. Model variants have different options and costs; the currently selected generator controls determine your request.
On this page
Before you begin
- An account with enough credits for the selected request.
- A short idea or a reference image appropriate for the selected model.
- A clear choice of the exact 2.0 model variant you intend to use.
1. Check which Seedance 2.0 model is selected
Open the video generator and inspect the model selector. Seedance 2.0, Seedance 2.0 Fast and Seedance 2.0 Mini are separate choices. Do not assume a homepage heading or a saved draft determines the exact option you will submit.
| Control | Why it matters |
|---|---|
| Full model name | Identifies the rate and capabilities used for this request. |
| Duration menu | Only the displayed supported durations can be submitted. |
| Resolution menu | A setting available on one variant may be absent on another. |
| Reference inputs | Different models can interpret or limit references differently. |
| Current credit estimate | Use this to compare the actual request. |
Use a short clip to explore an unfamiliar model. A lower quote does not establish that a variant will be adequate for your exact shot, and a higher quote does not guarantee a better creative result.
2. Pick text-to-video or image-to-video
Text-to-video gives you a blank starting point. Describe the subject, setting, action and camera so the model can construct the scene. Image-to-video starts from an existing visual reference; describe the intended movement and preserve the details that matter.
For the first image-led shot, use an object, landscape or original illustration with a clear subject. Observe the uploader's format and size limits. Real-person references can require additional review that this workflow does not provide, so do not begin with a personal photo and assume every model will accept it.
3. Write a short, actionable prompt
A single red toy train moves slowly along a wooden track beside a sunny window. The camera follows at the same speed from a low side view. The train keeps the same shape and color. Soft daylight, a tidy room, one continuous shot.
How to adapt itSelect the duration and output size in the generator. This template describes the shot, not a tested outcome.
Use the reference illustration for the character's appearance. The character gives a small wave with one hand and returns to a relaxed pose. Keep the outfit, face shape and colors consistent. The camera stays still and the illustrated style remains unchanged.
How to adapt itUse your own illustrated character. A single gesture is a useful first test before a longer sequence.
If the first prompt contains several scene changes, cut it down. You need to know whether the model follows the central action before adding more transitions, dialogue or camera instructions.
4. Set the output and review the quote
Choose duration, resolution and aspect ratio from the controls for the selected variant. A 5-second clip is a practical starting duration for a small action. Use 16:9 for a wide composition or 9:16 for a vertical one when the model offers them. The 2.0 routes do not provide the 30-second option covered in the 2.5 guide.
If the selected model exposes audio settings, check their state and the updated quote. Do not copy a cost from a different variant, resolution or sound configuration. Review all settings again if you switch models after writing the prompt.
5. Generate, review and keep your settings
- Submit one request after checking the quote and balance.
- Let the task reach a clear status before deciding whether to retry.
- Watch the full output, then inspect important frames for shape or continuity changes.
- Download a completed output and save its prompt and settings.
- Change one instruction for the next attempt, such as reducing camera movement.
If you need longer single generations, read the Seedance 2.5 guide and inspect its current controls. Choose between versions by testing the task you need, rather than by assuming the newer version is always the best fit.
Questions
Is Seedance 2.0 the same as Fast or Mini?
They appear as separate model options in SWRFlow. Rates, output options and behavior can differ. Always identify the full name in the selector and review the current duration, resolution and quote before generating or comparing costs.
Can I use a Seedance 2.5 prompt with 2.0?
A simple subject-and-action prompt can be adapted, but the timing and available inputs need to match the chosen 2.0 route. Remove unsupported instructions, shorten the event sequence and select the actual settings in the controls rather than relying on numbers in the prompt.
Do I need to use the Canvas?
You can begin in the standalone video generator. Canvas is useful later when you want to arrange references, model steps and outputs as a project. It is not required to follow the basic text-to-video or image-to-video workflow in this guide.
References & useful links
- ByteDance: Seedance 2.0 launch
Background on the model family.
- Runway: creating with Seedance 2.0
An example of the model in another platform. Its exact input limits and controls may differ.
Put the next step into practice
Check the full model name and use a short clip for your first attempt.
Choose a Seedance model