Seedance 2.0 Prompts: A Formula and 18 Examples

A good Seedance 2.0 prompt describes one shot in plain sentences: the subject, one clear action, one camera move, the light and style, and the sounds the viewer should hear, because Seedance 2.0 generates audio together with the video. Keep each clip to a single scene, draft at 4 seconds and 480p, and change one part of the prompt per attempt. Below are the formula, two real Kensa outputs with their exact prompts and settings, and 18 copy-ready prompts grouped by use.

What you can set for Seedance 2.0 on Kensa

Your prompt is sent to the model together with these settings, so pick them in the generator rather than writing them into the prompt.

Prompts and uploaded images are checked by content moderation before generation. Celebrities, children, brand logos, copyrighted characters and NSFW content are blocked, and you are not charged for a rejected request.

The Seedance 2.0 prompt formula

Write one shot, in this order:

1. Subject: who or what is on screen, with the details that matter ("a frosted glass serum bottle, unlabeled").
2. Action: one thing that happens, with a verb that shows pace ("a single drop falls and ripples outward").
3. Camera: one move or a static shot ("slow push-in", "low tracking shot", "static overhead shot", "slow orbit").
4. Style and light: time of day, light source, color, lens and genre ("soft top light with a warm rim light, macro lens, premium commercial style").
5. Sound: one ambient bed plus one or two sounds tied to what happens on screen ("a soft drip, a low room hum").

As a template:

[Subject] [does one action] in [setting]. [One camera move]. [Light, color and style]. Sound: [ambient sound], [sound tied to the action].

The first real prompt below already follows this order. Split into its parts:

Writing the sound cue

Seedance 2.0 makes sound effects and ambient sound that match the picture, and your prompt steers them. A few rules keep the cue useful:

• Name concrete sounds. "Rain on a tin roof" and "ice cracking in a glass" work better than "nice audio" or "immersive sound".
• Tie sounds to visible actions. A splash you can see is more likely to get a matching splash sound than a sound with no source on screen.
• Keep it to two or three sounds. One background bed and one or two effects are easier to get right than a long list.
• Leave out spoken lines. Kensa has no script or voice input for Seedance 2.0 and no lip sync. Add voiceover or licensed music in your editor.

If you plan to replace the sound anyway, turn the audio toggle off; the price is the same either way. The Seedance 2.0 audio guide covers sound prompts in more depth.

Real Seedance 2.0 outputs on Kensa

These two clips are real Kensa outputs, generated on October 31, 2026 for our monthly model benchmark. Both used Text to Video at 16:9, 8 seconds, 480p, with audio on. Each prompt was sent unchanged and run once, so treat them as examples of what the model does, not a guarantee.

Real output 1: puppy and soap bubbles

A golden retriever puppy chases soap bubbles across a sunlit backyard lawn, paws kicking up blades of grass, bubbles popping with soft sounds, low tracking camera alongside the puppy, warm afternoon light, cinematic.

• Settings: Seedance 2.0, Text to Video, 8 seconds, 480p, 16:9, audio on
• Output: 864×496, 8.1 seconds, with sound; the provider took 190 seconds to return it
• What came out: the puppy keeps its shape for the whole clip, bubbles drift through the frame and the camera stays low as asked. Rather than a sideways tracking shot, the camera pushes in while the puppy runs toward it. The picture is soft at 480p.

Real output 2: orange soda product shot

A glass bottle of sparkling orange soda stands on a wet black stone slab, ice cubes tumble around it in slow motion and droplets splash outward, studio lighting with a warm rim light, macro lens, premium commercial style.

• Settings: Seedance 2.0, Text to Video, 8 seconds, 480p, 16:9, audio on
• Output: 864×496, 8.1 seconds, with sound; the provider took 233 seconds to return it
• What came out: convincing splashes and falling ice, the drink reads as orange and the bottle has no invented label. The framing crops off the top of the bottle for the whole clip.

The soda prompt never says what must stay in frame. If you rerun it, add something like "the whole bottle in frame, centered". That is a suggested fix, not one we have tested.

Prompting with images

On Kensa, Seedance 2.0 takes images in two ways, and they need different prompts.

One image as the first frame (Image to Video)

On Image to Video, your image becomes the first frame of the clip. The model already sees the subject, colors and composition, so the prompt should describe only what changes: what moves, how the camera moves and what the viewer hears. Re-describing the photo adds nothing, and a description that contradicts it (a different color, a different background) makes the result drift. Products, illustrations and landscapes work best; prompt 17 below is an example.

Reference images (Reference to Video)

On Reference to Video, you upload one to three images of the same subject, such as a character, a mascot or a product, and the model composes a new scene around it. Refer to the subject through the images ("the plush fox mascot from the reference images") instead of describing its looks again, then write the scene, action, camera and sound as usual. Prompt 18 below is an example, and How to Keep a Character Consistent with Reference Images explains how to choose the images.

Seedance 2.0 accepts photos of adults you have permission to use. Photos of celebrities or children, and images with brand logos, are blocked for every model.

18 copy-ready Seedance 2.0 prompts

These are example prompts written to the formula above. Unlike the two clips above, they have not been run as tests, and results vary from run to run. The suggested length and aspect ratio go in the generator settings, not in the prompt.

Product ads

1. Serum bottle on wet slate (1:1 or 16:9, 5 seconds)

A frosted glass serum bottle with a gold dropper, unlabeled, stands on wet black slate. A single drop falls from the dropper and ripples across a shallow film of water around the base. Slow push-in, macro lens, shallow depth of field, soft top light with a warm rim light, premium commercial style. Sound: a soft water drip, a low room hum.

2. Earbuds case opening (16:9, 5 seconds)

A plain white wireless earbuds case on a matte concrete pedestal opens slowly, and the two earbuds lift slightly on a soft glow. The camera orbits a quarter turn around the case. Clean studio light, cool gray background, minimal product commercial. Sound: a soft magnetic click as the lid opens, a quiet electronic chime.

3. Running shoe splash (16:9 or 9:16, 5 seconds)

An unbranded white running shoe lands on a wet asphalt road at night and throws a spray of water outward in slow motion. Low static camera at ground level, neon reflections in the puddle, high-contrast cinematic lighting. Sound: a sharp splash, distant traffic hum.

4. Coffee pour (16:9, 6 seconds)

A matte ceramic mug on a wooden counter fills with dark coffee poured from a gooseneck kettle, steam curling upward. Slow push-in from a three-quarter angle, morning window light, warm brown tones, cozy café commercial style. Sound: coffee pouring, the kettle set down on wood, soft café ambience.

9:16 vertical clips for social feeds

5. Confetti box reveal (9:16, 4 seconds)

A plain cardboard box on a bright yellow backdrop bursts open and dozens of colorful paper confetti strips fly up toward the camera, then flutter down. Static camera, subject centered with empty space at the top, punchy saturated colors, bright even light. Sound: a sharp pop, paper rustling.

6. Soap cutting (9:16, 5 seconds)

A thin blade slices slowly through a block of lavender soap, and fine curls peel away and fall onto a white marble tray. Static overhead close-up, soft even daylight, pastel colors, crisp detail. Sound: a crisp scraping cut, soft crumbling.

7. Cat and feather toy (9:16, 6 seconds)

A ginger cat on a sunny windowsill swats at a dangling feather toy, misses, and tumbles onto a cushion below. Handheld camera at the cat's eye level, natural afternoon light, playful home-video feel. Sound: a soft thump on the cushion, birds chirping outside.

Cinematic shots

8. Lighthouse in a storm (21:9 or 16:9, 8 seconds)

A lone lighthouse on a rocky cliff at dusk as storm waves crash against the rocks below and its beam sweeps through falling rain. Slow aerial pull-back revealing the coastline, cold blue tones against the warm beam, widescreen film look. Sound: heavy surf, wind gusts, rain hitting stone.

9. Neon alley at night (21:9 or 16:9, 8 seconds)

A figure in a long coat walks away from the camera down a narrow rain-soaked alley lit by neon signs, steam rising from a street grate. Slow tracking shot from behind at shoulder height, teal and magenta reflections on wet pavement, film noir mood, light film grain. Sound: footsteps on wet stone, steady rain, a distant siren.

10. Desert highway (21:9 or 16:9, 10 seconds)

An old red pickup truck drives along an empty desert highway at golden hour, trailing a thin cloud of dust. Wide aerial shot that slowly descends and tracks alongside the truck, long shadows, warm orange light, 35mm film look. Sound: engine rumble, tires on gravel, steady wind.

Food

11. Ramen noodle lift (16:9 or 9:16, 5 seconds)

Chopsticks lift a tangle of noodles out of a steaming bowl of ramen on a dark wooden counter, broth dripping back into the bowl. Close-up, slow push-in, shallow depth of field, warm tungsten light, rich food commercial style. Sound: broth dripping, a busy kitchen clattering softly in the background.

12. Pizza cheese pull (9:16 or 1:1, 5 seconds)

A slice is lifted from a wood-fired pizza, a long string of melted cheese stretching beneath it, fresh basil leaves on top. Macro close-up, slow motion, static camera, warm glow from the oven behind, shallow depth of field. Sound: a crackling wood fire, the soft tear of crust.

13. Chocolate over ice cream (1:1 or 16:9, 6 seconds)

Warm chocolate sauce pours over a scoop of vanilla ice cream in a glass bowl and slowly runs down the sides as the ice cream starts to melt. Slow orbit around the bowl, dark moody background, soft key light from the side, indulgent dessert commercial. Sound: a thick pour, a spoon tapping glass.

Travel

14. Hillside village at sunset (16:9, 8 seconds)

A whitewashed hillside village above a calm blue sea at sunset, lights turning on in the windows one by one. Slow aerial push forward over the rooftops, warm pink and gold sky, travel film style. Sound: gentle waves far below, distant bells, evening wind.

15. Mountain railway (16:9 or 21:9, 10 seconds)

A red train winds along a mountain railway through snowy pine forests, clouds drifting below the peaks. Wide aerial tracking shot following the train, crisp morning light, blue and white tones, travel documentary style. Sound: train wheels clacking on the rails, wind over the ridge.

16. Night market walk (9:16 or 16:9, 8 seconds)

A busy night market street lined with food stalls under strings of warm lanterns, steam and smoke drifting from the grills, passersby in soft focus. Slow forward gimbal shot at walking pace, shallow depth of field, warm amber light, travel vlog feel. Sound: sizzling grills, the murmur of the crowd, a vendor's bell.

From your own images

17. First frame: a plant on a desk (Image to Video, ratio to match your photo, 5 seconds). Upload a photo of a potted plant on a desk by a window, then use:

The leaves sway gently as if a window just opened, sunlight shifts slowly across the desk and dust drifts through the light. The camera holds still, then pushes in slightly. Sound: a light breeze, birds outside, a distant street.

18. Reference images: a mascot on stage (Reference to Video, 16:9, 8 seconds). Upload one to three images of your mascot, then use:

The plush fox mascot from the reference images walks onto a small wooden stage, waves and takes a bow as paper confetti falls. Medium shot, slow push-in, warm spotlight on the mascot, soft blue background. Sound: light applause, confetti rustling.

Common mistakes and fixes

The Advanced Prompting Techniques post has more camera, lighting and style vocabulary that works across models.

How to iterate cheaply

Seedance 2.0 is billed per second: 7.7 credits/s at 480p and 16.9 credits/s at 720p, rounded up per clip. The cheapest setting is 4 seconds at 480p, which costs 31 credits, and the generator shows the exact cost of any other setting before you click Generate. Audio doesn't change the price.

1. Draft at 4 seconds and 480p. Check the subject, the action, the camera and the sound.
2. Change one part per attempt. If you change the camera and the light at the same time, you won't know which one fixed it.
3. Render the length you need. Once the short draft works, generate the clip at its final length. A longer clip is a new generation, so expect some variation.
4. Switch to 720p last. 720p costs 2.2 times as much per second as 480p, so use it only for a prompt you already trust.

If a generation fails, the credits reserved for it go back to your balance. When a prompt works, keep it as a base and swap one part at a time: same camera, light and sound, different subject. To test the same idea on another model, see How to Write Prompts for Veo 3.1.

Frequently asked questions

How long should a Seedance 2.0 prompt be?

Long enough to cover the five parts of the formula and no longer. Both real prompts on this page are under 40 words. Extra adjectives compete with the details that matter, so cut words that don't change the shot.

Can Seedance 2.0 generate dialogue or lip sync on Kensa?

No. On Kensa, Seedance 2.0 generates sound effects and ambient sound that match the picture. There is no script or voice input, no lip sync and no audio upload, so record voiceover separately and add it in your editor.

Does describing sound in the prompt cost extra?

No. Audio is generated at no extra cost, and the per-second price is the same with the audio toggle on or off.

Can I start from my own photo?

Yes. Image to Video takes one image as the first frame, and Reference to Video takes one to three images of the same subject (JPG, PNG or WEBP, up to 10MB each). Seedance 2.0 accepts photos of adults you have permission to use; celebrities, children, brand logos and NSFW content are blocked.

Should I put the aspect ratio or length in the prompt?

Set them in the generator; the prompt describes the shot. Do compose for the format, though: for 9:16, keep the subject centered and favor vertical motion, and for 21:9, describe wide scenery and sideways movement.

Try your prompt

Open the Seedance 2.0 video generator, keep audio on, set 4 seconds at 480p and paste one of the prompts above. For specs, pricing and how the model compares with others on Kensa, see the Seedance 2.0 model page and the Seedance 2.0 Complete Guide.