Seedance 2.0 prompts: pasteable templates and a real clip

A good clip starts with a good brief. This is a pasteable playbook of seedance 2.0 prompts you can copy into Omnely and run today: a 6-part formula, text-to-video templates, image-to-video prompts that use a start frame or a start and end frame, and a real Seedance 2.0 clip made on Omnely with the exact prompt shown. On Omnely, Seedance 2.0 takes text-to-video or image-to-video, makes clips of 4 seconds and longer, generates native audio by default, and accepts up to two reference images: one image is the start frame, two images are the start and end frames.

Every template below is written for that setup. Structure comes first, then the small tweaks that separate a jittery test clip from a shot worth publishing.

Anatomy of great seedance 2.0 prompts

Great seedance 2.0 prompts follow a repeatable 6-part formula. Write the pieces in this order and Seedance 2.0 renders motion, framing, and mood far more predictably than a wall of adjectives.

1. Subject — who or what is on screen, with one or two concrete appearance details.

2. Action — a single present-tense verb phrase describing visible motion.

3. Environment — location, time of day, weather, key set dressing.

4. Camera — one move only (dolly in, slow pan left, locked-off static, low-angle orbit, handheld push).

5. Style — lighting, film reference, color palette, texture.

6. Constraints — duration, aspect ratio, what to avoid.

Rules that pay off on Seedance 2.0

• One camera move per clip. Stacking "dolly in and pan and tilt up and orbit" produces mush. Pick one.
• Aim for 60–100 words in a single-shot prompt. Shorter prompts underspecify; longer prompts fight themselves.
• Lighting is the highest-return keyword group. "Golden hour rim light, soft haze, warm key from camera-left" changes a clip more than a pile of adjectives.
• Add one audio line. Native audio is on by default, so a short cue such as Audio: soft footsteps on wet pavement, distant traffic hum. shapes the ambience and Foley.
• Verbs beat adjectives. "Slowly turns toward the camera, exhales" outperforms "emotional."

The compact template

Subject: [who + 1–2 appearance details]
Action: [present-tense verb, one beat]
Environment: [location, time, weather]
Camera: [one move, for example slow dolly in]
Style: [lighting + film/color reference]
Constraints: [duration, aspect ratio, avoid list]
Audio: [optional ambience / Foley cue]

The same content also works as one flowing paragraph. The structured version is easier to iterate because only one line changes per run.

Seedance 2.0 prompts in action: a real clip from Omnely

The clip below was made on Omnely with Seedance 2.0: image-to-video, 16:9, 480p, 4 seconds, sound on. A glass perfume bottle image served as the start frame, so the prompt describes only motion, camera, light, and audio. Press play to hear the audio the prompt asked for.

Slow cinematic push-in on a faceted glass perfume bottle standing on wet black stone. Soft mist drifts across the stone, a few rose petals tremble in a light breeze, and the warm backlight slowly brightens behind the bottle. Audio: a soft ambient hum with one gentle glass chime. No text, no logos.

Notice what the prompt leaves out: nothing about the bottle's shape or color. The start frame already carries that, so every word goes to what changes over the four seconds.

Pasteable seedance 2.0 prompts: text-to-video

Copy any of these into the Seedance 2.0 generator. Each one fits a clip length of 4 to 12 seconds, and every prompt states duration and aspect ratio so the model does not drift.

1) Cinematic product hero (8s, 16:9)

Subject: A matte-black ceramic coffee mug with a thin gold rim, steam curling off the surface.
Action: Steam rises slowly and drifts to the right as the mug sits on a dark walnut counter.
Environment: Minimal café interior, early morning, soft window light from camera-left, out-of-focus copper espresso machine in the background.
Camera: Slow dolly in from medium shot to close-up, ending on the rim.
Style: Cinematic, shallow depth of field, warm color grade, 35mm film grain, gentle bokeh.
Constraints: 8 seconds, 16:9, no text, no logo, no hands in frame.
Audio: Faint milk-steamer hiss, distant jazz, muted café murmur.

2) Nature b-roll (6s, 21:9)

Subject: A single red maple leaf floating on the surface of a slow-moving forest stream.
Action: The leaf drifts left across frame while ripples move around it.
Environment: Autumn forest, late afternoon, dappled sunlight through evergreens, moss-covered stones underwater.
Camera: Locked-off overhead shot, static, no movement.
Style: Nature documentary, high dynamic range, saturated autumn palette, crisp water texture.
Constraints: 6 seconds, 21:9, keep the leaf centered vertically, no wildlife in frame.
Audio: Gentle stream flow, distant birdsong, light breeze in leaves.

3) Social ad: vertical product demo (5s, 9:16)

Subject: A pair of white minimalist wireless earbuds inside a matte charging case, sitting on a pastel-pink acrylic block.
Action: The case lid flips open smoothly and the earbuds glow with a soft LED pulse.
Environment: Bright studio, seamless pink-to-lavender gradient background, no props.
Camera: Slight orbit right, ending head-on.
Style: Clean e-commerce ad, high-key lighting, glossy reflections, punchy saturation, Reels-ready.
Constraints: 5 seconds, 9:16, no text overlay, no hands, keep the earbuds fully in frame throughout.
Audio: Soft magnetic snap, subtle synth sting on lid open.

4) Multi-beat mini-narrative (12s, 16:9)

When a prompt runs past 8 seconds, structure it as timed beats and keep it to a single continuous shot.

Subject: A young woman in a mustard-yellow raincoat, dark curly hair, holding a paper coffee cup.
Action:
0–4s — She stands under a shop awning watching heavy rain, breath visible.
4–8s — She steps out and starts walking left along a wet sidewalk, coffee steaming.
8–12s — She glances up and smiles slightly as sun breaks through clouds.
Environment: Narrow European street at dawn, cobblestones, glowing shop windows, wet reflections.
Camera: Slow handheld tracking left, keeping her in the right third of frame.
Style: Moody cinematic, teal-and-amber grade, subtle film grain, anamorphic-style flares.
Constraints: 12 seconds, 16:9, single continuous shot, no cuts, no other people in frame.
Audio: Rain on cobblestones, distant church bell, quiet footsteps.

5) Character action, dialogue-free (8s, 16:9)

Subject: A grizzled fisherman in his 60s, weathered face, wearing a navy wool sweater and yellow oilskin bib.
Action: He hauls a wet rope hand-over-hand, jaw set, eyes focused off-camera right.
Environment: Deck of a small fishing trawler at sea, overcast morning, cold spray in the air.
Camera: Low-angle medium shot, static, subtle boat sway.
Style: Documentary realism, desaturated palette, natural overcast light, sharp detail on skin and fabric texture.
Constraints: 8 seconds, 16:9, no dialogue, no visible fish, keep the rope taut throughout.
Audio: Wind, distant gulls, wet rope creaking, hull slap against waves.

Pasteable seedance 2.0 prompts: image-to-video and start/end frames

Image-to-video is where Seedance 2.0 shines for creators who already have a hero image. On Omnely, one reference image sets the start frame and two images set the start and end frames. Motion, camera, lighting shifts, and audio come from the text prompt.

The rule: describe motion only

Do not redescribe the still. Seedance already sees it. The prompt should say what changes between the first and last second.

6) Single reference (start frame): subtle product motion

Attach one hero product photo. This is the pattern behind the real clip above.

Motion: The liquid inside the perfume bottle shimmers and swirls gently, catching light on the glass facets. A soft plume of vapor drifts from the atomizer.
Camera: Very slow push-in, about 10% zoom over the full clip.
Lighting: Warm rim light from camera-right slowly intensifies.
Constraints: 5 seconds, 1:1, no hands, no text, keep the bottle centered.
Audio: Soft glass chime, faint atomizer hiss.

7) Two references (start + end): controlled transformation

Attach image 1 (closed laptop on a desk) and image 2 (the same laptop open with a glowing screen).

Motion: The laptop lid rises smoothly from fully closed to fully open, screen fading up from black to a soft blue glow.
Camera: Static locked-off three-quarter shot.
Lighting: Ambient desk lamp warms slightly as the screen glow appears.
Constraints: 4 seconds, 16:9, no hands, no reflections of a person.
Audio: Quiet hinge, soft startup chime.

8) Two references: a character walks into a new environment

Attach image 1 (a character standing indoors near a doorway) and image 2 (the same character silhouetted outside on a sunlit path).

Motion: The character walks forward through the doorway, moving from interior shadow to bright exterior sunlight.
Camera: Slow dolly in behind the character, matching their pace.
Lighting: Cool interior fades to warm golden-hour exterior across the clip.
Constraints: 6 seconds, 16:9, keep character identity and clothing consistent, no other people.
Audio: Footsteps on wood, then gravel, birds emerging as the character steps outside.

Timeline and multi-shot seedance 2.0 prompts

Each Seedance 2.0 clip is one continuous shot. To suggest several beats inside one clip, write timestamp blocks with a subject anchor that never changes, and pick a camera move that can motivate the beat changes (a slow push-in, an orbit, or a handheld drift). For a genuine multi-shot edit, generate several clips and cut them together in an editor.

9) Shot-list prompt for a 3-clip sequence

Run each shot as its own Seedance 2.0 generation. Keep the subject anchor identical across all three to preserve continuity.

SUBJECT ANCHOR (paste into every shot):
A tall barista, mid-20s, close-cropped black hair, small silver hoop earring, wearing a charcoal apron over a white tee.

SHOT 1 — Establishing (6s, 16:9)
Action: The barista wipes down the espresso bar, steam rising from the machine behind.
Camera: Slow dolly in from wide to medium.
Style: Warm morning light, film grain, cinematic.
Audio: Espresso machine hiss, quiet indie track underneath.

SHOT 2 — Action close-up (5s, 16:9)
Action: The barista tamps espresso grounds into a portafilter, hands only in frame.
Camera: Overhead top-down, static.
Style: Crisp macro detail, warm key light.
Audio: Sharp tamp click, low café ambience.

SHOT 3 — Reveal (6s, 16:9)
Action: The barista slides a finished latte across the counter and smiles slightly.
Camera: Slow rise from cup to face.
Style: Same warm grade, soft bokeh in background.
Audio: Ceramic on wood, soft café ambience.

10) Single-clip timeline (advanced, 12s)

Subject: A vintage 1970s orange Porsche, chrome trim, sitting on a coastal cliff road.
Action:
0–3s — The car sits still, engine idling, exhaust visible.
3–8s — The car pulls away and drives along the winding road toward the horizon.
8–12s — The car disappears around a bend, dust settling.
Environment: Rugged coastline, late afternoon, ocean fog rolling in.
Camera: Slow crane up from ground level to high wide.
Style: Cinematic, teal-and-orange grade, anamorphic feel.
Constraints: 12 seconds, 21:9, single continuous shot.
Audio: Air-cooled flat-six rumble, gulls, ocean below.

Prompt mistakes that ruin Seedance 2.0 output

These failure modes show up in almost every disappointing clip.

• Stacked camera moves. "Dolly in, then pan right, then tilt up, then orbit" tells Seedance nothing useful. Pick one primary move. If two are needed, make one dominant ("slow dolly in, with very slight handheld drift").
• The word "fast." It tends to read as motion smear and warping. Use "brisk," "purposeful," or "quick step," or say what moves fast ("camera stays static as leaves blow quickly across frame").
• Too many beats in a long clip. Three beats in 12 seconds works. Six beats in 12 seconds produces morphing. If the story needs six beats, cut it into several clips with the shot-list method above.
• Redescribing an image-to-video still. With an image attached, spend the words on motion, camera, lighting shift, and audio. Seedance sees the reference.
• No aspect ratio. State 16:9, 9:16, 21:9, or 1:1 in the constraints line so the framing does not drift.

A sensible loop: test the prompt at 480p, lock the wording, then re-run the keeper at a higher resolution.

Run these seedance 2.0 prompts on Omnely

Every prompt above runs on Seedance 2.0 inside Omnely, an all-in-one workspace for AI models with image, video, and chat in one login.

• Open the model page: Seedance 2.0 on Omnely
• Compare another video engine in the same workspace: Kling 3.0
• Step-by-step walkthrough: How to use Seedance 2
• Weighing models first? Read Seedance vs Kling
• Working from a photo? The Kling image-to-video guide covers reference-image tactics that transfer to Seedance 2.0.

Paste a prompt into the Seedance 2.0 generator, pick the aspect ratio and duration, add up to two reference images for start and end frame control, and generate. Omnely's video models need a paid plan, and new users get free credits at sign-up.