Seedance 2.0 Prompts: A Copy-Paste Library That Lands

Most Seedance 2.0 clips fail for one reason: the prompt describes a picture, not a shot. Seedance 2.0 builds motion, audio, and multiple camera angles in a single pass — so a static, image-style prompt wastes what the model does best.

This is a working Seedance 2.0 prompts library. Every block below is copy-paste ready, built around the model's real levers: multimodal references, the "lens switch" multi-shot cue, native audio, start-and-end frames, and a locked-off camera. You can run these on Seedance 2.0 directly — the model is live in ZenCreator's Video Generator at 480p and 720p, in clips from 5 to 15 seconds. New to it? Start on the uncensored Seedance 2.0 landing.

TL;DR: Structure each prompt as Subject → Action → Setting → Camera → Lighting → Audio. Add lens switch for multiple shots in one clip, address up to 9 images / 3 videos / 3 audio files as [Image N] / [Video N] / [Audio N], and describe the sound out loud so the native audio track has something to render. Copy the templates below and change the nouns.

What makes a Seedance 2.0 prompt actually land?

A prompt lands when it describes change over time — action and camera — not just a frozen scene. Seedance 2.0 animates and scores audio in one pass, so motion and sound have to be written in.

The failure mode is writing a photo caption: "A woman in a red dress on a rooftop." The model gets no verb, no camera intent, and no sound cue, so it guesses. A prompt that names the motion, the camera move, and the audio gives the model a job instead of a riddle.

Six fields cover everything the model needs:

• Subject — who or what, with 2–3 concrete details (hair, wardrobe, material)
• Action — a real verb: turns, sips, sprints, dissolves — not "is standing"
• Setting — where it happens, plus time of day
• Camera — one move per clip: static, slow pan, dolly in, orbit, handheld
• Lighting / mood — light source and color, e.g. "cool blue window light, warm sodium lamp"
• Audio — say the sound out loud: dialogue, ambient, sound effects, music

Search interest confirms people are hunting for exactly this — queries for Seedance prompt variants have climbed through 2026, with "seedance vs kling" up 136% quarter-over-quarter and "seedance 2.0 unrestricted" up 600%^[3].

What is the copy-paste prompt formula for Seedance 2.0?

Use one fenced template and swap the bracketed parts. This is the backbone every prompt example below is built on.

[Subject with 2–3 concrete details] [action verb, present tense] in [setting], [time of day].
[Camera move: static / slow pan / dolly in / orbit / handheld].
[Light source and color]. [Audio: dialogue, ambient sound, SFX, music].

Worked example — the exact prompt we ran to produce the clip below:

A woman with auburn hair in a charcoal wool coat stands at a rain-streaked
train-station window at dusk, steam rising outside. She lifts a paper coffee cup
and takes a slow sip, eyes drifting to the platform. Lens switch to a close-up of
her reflection in the wet glass as a train blurs past behind her. Cool blue
ambient light, warm sodium lamp glow. Ambient station announcements, soft rain,
distant train rumble.

We ran this on ZenCreator's Seedance 2.0 (seedance_2_0, 720p, 5s, native audio on). Notice it already uses a lens switch — the model cuts from the wide shot to the reflection close-up inside a single 5-second clip.

Real Seedance 2.0 output on ZenCreator — text-to-video, 720p, 5s, native audio. The prompt above generated this exact clip.

How do you write a cinematic Seedance 2.0 prompt?

Name a lens and a single camera move, then commit the light to one direction and color. Cinematic look comes from restraint, not from stacking adjectives.

Cinematic portrait:

A man in his 40s with a grey stubble beard and a navy overcoat stands on a
foggy pier at dawn. He turns slowly toward camera, breath visible in the cold.
Shallow depth of field, 85mm lens, static shot with subtle handheld sway.
Soft grey backlight through fog, cold blue palette. Gentle wave lapping,
distant foghorn, wind.

Product shot:

A matte-black mechanical watch rests on wet slate stone. The camera orbits
90 degrees around it as a single hard key light rakes across the dial.
Deep shadows, cool reflections. Low ambient hum, faint ticking.

Two rules do most of the work here: pick one camera move per clip (an orbit and a dolly and a pan in five seconds looks broken), and commit the light to a single source and color instead of "beautiful lighting."

How do multimodal reference prompts work in Seedance 2.0?

Seedance 2.0 accepts up to 9 reference images, 3 reference videos, and 3 audio files, and you address them in the prompt as [Image N], [Video N], and [Audio N]^[1]. This is the model's signature feature — it turns a prompt into a director's brief.

In ZenCreator's Video Generator, you upload the references, then reference them by number. Each slot has a job:

• Image slots — a face for consistency, a product, a background plate, a wardrobe reference
• Video slots — a motion or camera move to copy (≤15s total across all 3)
• Audio slots — a voice, a music bed, or an ambient track (≤15s total across all 3)

Character consistency + custom background:

[Image 1] is the main character. Place her in the environment from [Image 2],
walking slowly toward camera through the doorway. Keep her face and hairstyle
from [Image 1] exact. Warm interior light, evening. Footsteps on wood, low music.

Motion transfer from a reference video:

Apply the camera movement and pacing of [Video 1] to the character in [Image 1].
She performs the same walk-and-turn. Neon-lit street at night, rain reflections.
Copy the rhythm of [Video 1] exactly.

Audio-driven scene:

[Image 1] lip-syncs the dialogue in [Audio 1] while seated at a cafe table.
Natural facial expression and head movement matched to the speech.
Soft daylight from a window on the left. Use [Audio 1] as the voice track.

The reference system is why creators pick Seedance 2.0 over prompt-only models — you lock a face, a location, and a voice instead of re-rolling until the character stops drifting.

How do you get multiple shots in one clip with "lens switch"?

Write the word lens switch between two described shots and Seedance 2.0 cuts to a new angle while keeping the subject consistent. It packs a mini-sequence into one generation.

Most models give you a single continuous take. lens switch lets you storyboard a cut — wide to close-up, or over-the-shoulder to reverse — inside a single 10–15 second clip.

Two-shot reveal:

Wide shot of a detective in a trench coat approaching an abandoned warehouse
door at night. Lens switch to a close-up of his hand pushing the door open.
Lens switch to a low-angle shot of a single bulb swinging inside.
Tense low strings build throughout, creaking hinge, echoing footsteps.

Product-to-reaction cut:

Close-up of espresso pouring into a white cup on a marble counter, steam rising.
Lens switch to a woman's face lighting up as she takes the first sip.
Warm morning light. Pouring liquid, a soft sigh, quiet cafe murmur.

Give longer clips (12–15s) two or three lens switch beats so the model has time to hold each shot. On a 5-second clip, one cut is plenty.

How do you add native audio, start-and-end frames, and a fixed camera?

Turn on native audio and describe the sound in the prompt; set a start and end frame to control the arc; enable fixed camera to lock the frame for product or talking-head shots.

Native audio. Seedance 2.0 scores the video in the same pass — dialogue, sound effects, ambient, and music^[1]. It only renders what you name, so end every prompt with a sound line. For dialogue, write it in quotes:

A barista leans over the counter and says, "Your usual, right?" with a warm
smile. Cozy cafe, morning light. Her voice, espresso machine hiss, low chatter.

Start-and-end frame. When you supply both a first and a last frame, the model interpolates the whole motion between them — ideal for a controlled transformation (closed door → open door, day → night). In the Video Generator, set the starting image and the target final frame, then describe the transition:

Start on the closed antique door. Over the clip it swings fully open to reveal
a candlelit library. Slow push-in. Creaking wood, a low ambient hum.

Fixed camera. Toggle fixed camera (available on seedance_2_0, not the fast variant) when you want zero drift — product turntables, locked-off talking heads, before-and-after shots. Then describe subject motion only:

A locked-off shot of a perfume bottle on a rotating white pedestal.
The bottle turns; the camera never moves. Clean studio light, soft reflections.
Faint ambient tone.

If you are new to the model, the how to use Seedance 2.0 guide walks through the interface, resolutions, and durations before you start pasting prompts.

Video templates to start from

These are popular ZenCreator video templates — general Video Generator presets, not Seedance-specific ones. A fast way to see a working clip with the prompt already loaded; switch the engine to Seedance 2.0 to run any of them on this model.

For more model background and prompt theory across engines, see the WAN video complete guide — the same Subject → Action → Camera → Audio structure carries over to every video model in ZenCreator.

Frequently Asked Questions

What is the best prompt structure for Seedance 2.0?

Subject → Action → Setting → Camera → Lighting → Audio. Lead with a concrete subject, give it a real action verb, name one camera move, commit the light to a single source and color, and describe the sound out loud. This structure covers every Seedance prompt example in this guide.

How many reference files can a Seedance 2.0 prompt use?

Up to 9 images, 3 videos, and 3 audio files per generation. You address them in the prompt as [Image N], [Video N], and [Audio N]. Videos and audio are each capped at 15 seconds total across their slots.

What does "lens switch" do in a Seedance 2.0 prompt?

It signals a cut. Writing lens switch between two described shots tells the model to change camera angle while keeping the subject consistent, so you get a multi-shot sequence inside one clip. Give 12–15 second clips two or three switches; use one on a 5-second clip.

How do I get audio in a Seedance 2.0 clip?

Turn on native audio and describe the sound in the prompt. The model scores dialogue, sound effects, ambient noise, and music in the same pass, but only renders what you name — so end every prompt with a sound line, and put spoken lines in quotes.

What resolution and length can Seedance 2.0 generate on ZenCreator?

480p or 720p, in durations of 5, 6, 8, 10, 12, or 15 seconds. Both text-to-video and image-to-video are available, with native audio, fixed camera, start-and-end frames, and the multimodal reference system.

Where can I get a Seedance 2.0 prompt guide with real examples?

This page is a working Seedance 2.0 prompt guide — every template is copy-paste ready and built on the model's real features. Paste any block into ZenCreator's Video Generator, swap the nouns, and generate.