Seedance Prompts
24 copy-paste prompts for Seedance 2.0, ByteDance's native-audio AI video model, built around what it does best: text-to-video with sound, image-to-video using the @ reference system, multi-shot stories, and cinematic camera control.
Last updated July 17, 2026
What Seedance 2.0 Is & How to Use It
Seedance 2.0 is the AI video generation model from ByteDance's Seed research team, launched in February 2026. It is built on a unified multimodal architecture, meaning a single model understands text, images, audio, and video together. That design is why it can do the thing it is best known for: generate native audio and video in one pass, with synchronized dialogue, ambient sound, and music, rather than bolting audio on afterward. As of 2026 it trades the top spot on the Artificial Analysis leaderboards with Kling 3.0 and Veo 3.1.
For text-to-video, you simply describe the scene and the sound, and Seedance produces both. For image-to-video, it adds an @ mention reference system: you upload assets and point to each one in the prompt so the model knows its job, @Image1 as the first frame, @Video1 for camera movement, @Audio1 for background music. This explicit control is what lets Seedance preserve your exact subject and composition while adding physically accurate motion. It also supports multi-shot storytelling from a single prompt, so one prompt can describe two or three beats that share a continuous audio timeline.
Videos generate at up to 2K resolution in 16:9, 4:3, 1:1, 3:4, or 9:16, with durations of roughly 4 to 15 seconds. Short clips with one clear action look best. The prompts below are grouped around exactly these strengths, native-audio text-to-video, @-referenced image-to-video, multi-shot sequences, cinematic camera control, and marketing shorts. Structure your own prompts in the order subject and action, scene, lighting, camera, then a short audio note for the most reliable results.
24 Copy-Paste Seedance Prompts
Grouped by what Seedance 2.0 does best. Copy a block, attach your image or audio where the @ mentions call for it, and paste into your Seedance surface.
Text-to-Video With Native Audio
Seedance's signature move: describe the scene AND the sound in one prompt, and it generates both together. End each prompt with an audio note.
A barista pulls a shot of espresso on a matte-black machine in a cozy morning cafe. Warm window light, shallow depth of field, slow push-in on the portafilter. Audio: the hiss of steam, a soft grind, and low ambient chatter.
A wave curls and breaks on an empty beach at golden hour, filmed in slow motion from a low angle just above the water. Warm backlight, fine sea spray catching the sun. Audio: the deep rush of the wave and distant gulls.
A chef plates a fine-dining dessert: a spoon drags a streak of raspberry coulis across a white plate, then places a quenelle of ice cream. Top-down 45-degree angle, soft studio light. Audio: gentle clink of the spoon and quiet kitchen ambience.
A young woman in a yellow raincoat walks through a neon-lit Tokyo alley at night, reflections shimmering in puddles. Handheld tracking shot behind her. Audio: light rain, distant traffic, and one line of dialogue: 'Almost there.'
A cat stretches and yawns on a sunlit windowsill, then curls back to sleep. Close-up, warm afternoon light, very subtle camera drift. Audio: a soft purr and birdsong through an open window.
Image-to-Video With the @ Reference System
Upload a reference image and drive motion with @ mentions. Seedance keeps your subject and composition steady while adding realistic movement.
@Image1 as the first frame. Bring the portrait to life: the subject slowly turns their head toward the camera and smiles, hair moving gently in a light breeze. Keep the exact face, outfit, and background. Soft natural light. Audio: quiet ambient room tone.
@Image1 as the first frame (a product photo of a sneaker on a pedestal). Slowly orbit the camera 180 degrees around the sneaker with a subtle rim light sweep, keeping the shoe's shape, colorway, and logo identical. Audio: a soft cinematic synth pad.
@Image1 as the first frame. Animate this landscape: drifting clouds, rippling water, and swaying grass, while keeping the composition and color grade unchanged. Slow, steady dolly forward. Audio: gentle wind and flowing water.
@Image1 as the first frame and @Audio1 for background music. Turn this band photo into a stylized performance clip: the musicians play in sync, stage lights pulse to the beat, light haze in the air. Preserve each person's appearance.
@Image1 as the first frame, @Video1 for camera movement. Apply the sweeping crane motion from @Video1 to my scene in @Image1, keeping my subject and setting intact while matching that camera path. Audio: cinematic ambience.
Multi-Shot Storytelling
Describe two or three beats in order and Seedance stitches them into one short sequence with a shared audio timeline. Number the beats.
A three-beat morning routine, one continuous mood. Beat 1: an alarm clock on a nightstand as a hand reaches to silence it, dim dawn light. Beat 2: coffee dripping into a glass carafe, warm kitchen light. Beat 3: a front door opening to a bright street. Audio: alarm, coffee drip, then city morning.
A short product story in three shots. Beat 1: a smartphone box on a desk, top-down. Beat 2: hands lifting the lid, soft magnetic click. Beat 3: the phone held up to catch window light, screen glowing. Consistent warm lighting throughout. Audio: subtle foley and a light synth.
A two-beat travel teaser. Beat 1: a plane wing above a sea of clouds at sunrise, wide shot. Beat 2: a traveler steps onto a sunlit cobblestone street with a rolling suitcase. Cinematic color grade. Audio: airy pad, then bustling street ambience.
A three-beat cooking sequence. Beat 1: fresh vegetables being chopped on a wooden board, close-up. Beat 2: they sizzle as they hit a hot pan, steam rising. Beat 3: the finished dish plated and garnished. Warm kitchen light. Audio: chopping, sizzling, then quiet.
A two-beat character intro. Beat 1: a detective in a trench coat steps out of the rain into a dim doorway, backlit. Beat 2: close-up as they look up, neon reflected in their eyes. Moody noir lighting. Audio: rain, a distant siren, one line: 'Let's begin.'
Cinematic Camera & Lighting Control
Seedance gives director-level control. Name the shot, the movement, and the light explicitly for a polished, cinematic look.
A lone lighthouse on a cliff during a storm. Start on a wide establishing shot, then a slow crane up to reveal the churning sea below. Dramatic overcast light with breaks of sun. Audio: crashing waves, wind, and a low orchestral swell.
A luxury watch on black velvet. Extreme macro, slow focus pull from the crown to the dial, a single moving key light creating a highlight sweep across the metal. Audio: a minimal, elegant tick and soft ambient tone.
A dancer in a bare studio. Orbiting camera at mid-height circling the dancer as they move through a slow, expressive sequence, hard directional light casting long shadows. Audio: a single cello line matched to the movement.
A city skyline at blue hour. A smooth aerial dolly gliding between skyscrapers as office lights flicker on, cool cinematic grade with warm window accents. Audio: a gentle ambient drone and faint distant traffic.
A close-up of a face lit only by a phone screen in the dark. Slow push-in, cool blue key light, soft falloff into shadow, subtle catchlights in the eyes. Audio: quiet room tone and a single notification chime.
Marketing, Ads & Social
Turn a product or concept into a scroll-stopping short. Ask for 9:16 for vertical social, and keep the brand element consistent.
A vertical 9:16 ad for a skincare serum. The bottle rises out of soft ripples of water in slow motion, droplets sliding down the glass, clean bright studio light, seamless white background with space for text at the top. Audio: a calm, premium ambient tone.
A 9:16 energy-drink hero shot: the can slams onto a wet concrete surface, splash frozen mid-air, bold contrasty light, subtle lens flare. Audio: a punchy impact and an electric riser.
A cozy 9:16 lifestyle clip for a candle brand: a hand lights the wick, warm flame flickers, soft evening light in a styled living room, shallow depth of field. Audio: a gentle crackle and quiet acoustic guitar.
A 16:9 SaaS explainer opener: an abstract flow of glowing data nodes connecting across a dark gradient, smooth camera glide, brand-blue accent light. Audio: a clean, modern synth pulse. Leave the center clear for a logo.
Give your Seedance videos a studio-grade voiceover
Seedance nails the visuals and ambient sound; for polished narration or a branded voice, ElevenLabs turns your script into lifelike speech in 70+ languages and can clone your own voice for a consistent channel. Free to start.
Affiliate link, we may earn a commission at no extra cost to you.
8 Tips for Better Seedance Results
- Structure every prompt as: subject and action, then scene, then lighting, then camera, then audio. Seedance rewards that order.
- Always add an audio note; native sound is Seedance's edge, so 'Audio: soft rain and one line of dialogue' beats leaving it silent.
- Keep clips 4 to 12 seconds with one clear action. Overloaded prompts with many simultaneous events reduce reliability.
- For image-to-video, use @ mentions: '@Image1 as the first frame', '@Video1 for camera movement', '@Audio1 for music'.
- For sequences, number the beats ('Beat 1... Beat 2...') so the model paces the shots and shares one audio timeline.
- Name the exact camera move (crane up, slow dolly in, 180-degree orbit) instead of vague terms like 'cinematic'.
- Request 9:16 for TikTok, Reels, and Shorts; 16:9 for YouTube and widescreen; state it explicitly in the prompt.
- For lip-sync, give a short line of dialogue in the target language; Seedance supports phoneme-level sync in 8+ languages.
Turn your AI videos into a real website or app
Building a portfolio, an AI-video tool, or a client gallery for your Seedance work? Lovable builds the whole site from a plain-English prompt, no code. Free to start.
Affiliate link, we may earn a commission at no extra cost to you.
Seedance FAQ
What is Seedance 2.0?
Seedance 2.0 is ByteDance's AI video generation model, built by its Seed research team and launched in February 2026. It uses a unified multimodal architecture that accepts text, image, audio, and video inputs, and it was the first widely available model to generate native audio and video together in a single pass rather than adding sound afterward. As of 2026 it sits at the top of the Artificial Analysis video leaderboards alongside Kling 3.0 and Veo 3.1.
What makes Seedance different from other AI video models?
Three things stand out. First, native audio-video generation: dialogue, ambient sound, and music are produced inside the same generation, with phoneme-level lip-sync in 8+ languages. Second, multi-shot storytelling from a single prompt, so one prompt can describe a short sequence of beats. Third, the @ mention reference system for image-to-video, which lets you tell the model exactly how each uploaded file is used (first frame, camera reference, or audio track). Together these make it strong for short narrative clips, not just single shots.
How do I write a good Seedance prompt?
Seedance follows natural language closely, so structure the prompt around WHO performs WHAT action, then the spatial background, then lighting and visual style, then camera choreography, and finally an audio note. A clear pattern is: subject and action, scene and background, lighting and mood, camera movement, then sound (ambient, music, or one line of dialogue). Keep clips between 4 and 12 seconds with simple, clear motion for the most reliable results, and describe two or three beats in order for multi-shot sequences.
How does the @ reference system work?
When you upload assets to Seedance 2.0, you point to each one in your prompt with an @ mention so the model knows its role. For example, write '@Image1 as the first frame' to start the video from your reference image, '@Video1 for camera movement' to borrow motion from a clip, and '@Audio1 for background music' to set the soundtrack. This explicit control is what makes Seedance's image-to-video mode preserve your subject and composition while adding physically accurate motion.
What resolution and length can Seedance produce?
Seedance 2.0 generates video at up to 2K resolution in 16:9, 4:3, 1:1, 3:4, or 9:16 aspect ratios, with durations from about 4 to 15 seconds per generation. Short clips (4 to 12 seconds) with clear actions tend to look best. For vertical social formats, ask for 9:16; for cinematic widescreen, use 16:9. You can chain multiple generations to build longer sequences.
Is Seedance free, and how do I access it?
Access is available through ByteDance's own surfaces (including Dreamina) and via third-party model hubs and APIs such as BytePlus ModelArk, WaveSpeed, and others that offer pay-as-you-go Seedance generations. Free trial credits are common on these hubs, but sustained or commercial use is billed per generation and varies by provider. Check the specific platform for current pricing and any free daily quota before scaling up. This page's prompts work across any surface running Seedance 2.0.