This is a reference sheet for prompting Seedance, ByteDance's text-image-audio-to-video model. Bookmark it, and when a clip misses, scan the tables for the exact keyword that fixes it — the shot size, the camera move, the lighting, or the audio line.
The rule underneath all of it: describe the shot the way a director would. Seedance 2.0 (February 2026) and the newer Seedance 2.5 (announced June 2026) generate picture and sound in a single pass, so every prompt names a subject, a move, a light, and — when you want it — the audio.
The prompt formula
Every strong Seedance prompt follows one order: Subject + Action + Environment + Camera move & framing + Lighting + Style, then optional Audio and a settings note. Write it as one flowing paragraph, use positive phrasing, and end with the settings reminder in parentheses.
| Part | What it sets | Example |
|---|---|---|
| Subject | Who or what is on screen | "A lone figure in a wet trench coat" |
| Action | What they do | "walks slowly, hands in pockets" |
| Environment | Setting and time | "a narrow neon-lit alley at night" |
| Camera | Move and framing | "medium wide shot, slow dolly-in" |
| Lighting | Mood and source | "low-key, practical neon, anamorphic flares" |
| Style | Look and grade | "cinematic, shallow depth of field" |
| Audio | Speech, SFX, ambient | "Ambient: steady rain, distant traffic" |
| Settings | Ratio, resolution, length | "(16:9, 2K, 8s)" |
A lone figure in a wet trench coat walks slowly down a narrow neon-lit alley at night, hands in pockets, breath visible in the cold. Rain-slicked cobblestones reflect pink and cyan signage. Medium wide shot, slow dolly-in following behind the figure. Low-key lighting, practical neon, anamorphic flares, shallow depth of field. Ambient: steady rain, distant traffic, a dripping gutter. SFX: footsteps splashing through puddles. No subtitles, no on-screen text. (16:9, 2K, 8s)Want the reasoning behind each part? Our guide on how to prompt Seedance for realistic video walks through it, and the 35 best Seedance prompts roundup shows the formula in finished form.
Shot sizes
Shot size sets how much of the subject the frame holds and how close the viewer feels. Name it right after the action so Seedance frames before it moves.
| Shot | Keywords | When to use |
|---|---|---|
| Wide / establishing | "wide shot", "establishing shot", "extreme wide" | Set the location and scale at the top of a scene. |
| Full shot | "full shot", "full-body shot" | Show the whole subject and their movement or wardrobe. |
| Medium | "medium shot", "waist-up", "cowboy shot" | Everyday framing for action and one-person dialogue. |
| Medium close-up | "medium close-up", "chest-up" | Conversation with a little more emotion than a medium. |
| Close-up | "close-up", "tight on the face" | Emotion, reactions, and lip-sync dialogue lines. |
| Extreme close-up | "extreme close-up", "insert shot" | Eyes, hands, a detail — a beat of tension or focus. |
| Macro | "macro shot", "macro detail" | Product textures, water droplets, tiny mechanical detail. |
| Over-the-shoulder | "over-the-shoulder shot", "OTS" | Two-person dialogue and point-of-view exchanges. |
Camera moves
One camera move per short clip keeps the motion clean. Pick the move that matches the emotion, and use the exact keyword — Seedance reads these terms directly.
| Move | Keywords | Effect |
|---|---|---|
| Push in | "slow dolly-in", "push in", "camera moves closer" | Builds intensity and draws the eye to the subject. |
| Pull back | "pull back", "dolly out", "reveal shot" | Reveals context or isolates the subject in a space. |
| Orbit / arc | "orbit around the subject", "arc shot", "360 orbit" | Adds energy and shows a subject from all sides. |
| Tracking | "tracking shot", "camera follows behind", "dolly alongside" | Moves with a walking or running subject. |
| Pan / tilt | "slow pan left", "whip pan", "tilt up" | Sweeps across a scene or connects two points. |
| Bird's eye / drone | "bird's-eye view", "aerial drone shot", "top-down" | Scale, landscapes, and dramatic overhead reveals. |
| Low angle | "low-angle shot", "worm's-eye view" | Makes the subject look powerful or looming. |
| Static / locked | "static shot", "locked-off camera", "tripod shot" | Calm, composed frames and clean dialogue. |
| Handheld | "handheld camera", "shaky documentary style" | Raw, urgent, real-feeling energy. |
Lighting
Lighting carries the mood more than any other keyword. Name the source and the quality, and pair it with a time of day when it helps.
| Lighting | Keywords | Mood it gives |
|---|---|---|
| Golden hour | "golden hour", "warm sunset light", "long shadows" | Warm, nostalgic, flattering on faces. |
| Blue hour | "blue hour", "cool twilight", "dusk glow" | Calm, cinematic, slightly melancholy. |
| Low-key | "low-key lighting", "deep shadows", "single hard light" | Moody, dramatic, noir and thriller looks. |
| High-key | "high-key lighting", "bright and even", "soft shadows" | Clean, upbeat, commercial and lifestyle. |
| Studio three-point | "three-point lighting", "key, fill and rim light" | Professional interviews and product hero shots. |
| Practical / neon | "practical neon", "practical lamps", "signage glow" | Urban night, cyberpunk, music-video energy. |
| Natural / overcast | "soft natural light", "overcast diffused light" | Documentary realism and gentle skin tones. |
| Backlit / silhouette | "backlit", "rim light", "silhouette against the window" | Mystery, shape, and a glowing edge. |
| Volumetric | "volumetric light", "god rays", "haze and light shafts" | Atmosphere and depth in dusty or foggy air. |
Native audio syntax
Seedance generates audio in the same pass as the picture, so you write sound in plain words. Use these labels so speech, effects, ambience, and music stay separate and clean.
| Element | How to write it | Notes |
|---|---|---|
| Dialogue | She says, "We should have left an hour ago." | Quotes trigger automatic phoneme-level lip-sync in 8+ languages. |
| Sound effects | SFX: keys jingling, a door creaking open | List one or two specific effects tied to the action. |
| Ambient | Ambient: steady rain, distant traffic, low wind | Sets the background bed for the whole scene. |
| Music mood | Music: slow, tense synth pulse that builds | Describe mood and tempo, not a named song. |
| Track sync | Sync the cut to the beat of @Audio1. | @Audio1 supplies a music, rhythm, or dialogue track to match. |
| Captions off | No subtitles, no on-screen text. | Add this whenever dialogue or graphics could bleed captions. |
References & multi-shot
Multimodal @mentions lock in a face, a style, a motion, or a track, and time-coded segments let one prompt cover several shots. Write the references into the sentence, and keep the subject consistent across segments.
| Reference | What it supplies | How to write it |
|---|---|---|
| @Image1 | Character face and appearance | Use @Image1 as the character's appearance. |
| @Image2 | Style and aesthetic | Match the color and style of @Image2. |
| @Video1 | Camera-movement or motion reference | Copy the camera motion from @Video1. |
| @Audio1 | Music, rhythm, or dialogue track | Sync the action to the beat of @Audio1. |
| Time-coded segments | Multi-shot storytelling in one clip | 0-3s: wide shot… 3-6s: medium… 6-10s: close-up. |
For fill-in-the-blank skeletons that use these references, grab our Seedance prompt templates.
Settings
Aspect ratio, resolution, and duration are app or API controls in Dreamina, fal.ai, and similar tools — not inline flags. There is no --ar like Midjourney, so set the controls to match and add a plain reminder in parentheses at the end of the prompt.
| Setting | Options | When to use |
|---|---|---|
| Aspect ratio | 16:9, 9:16, 1:1, adaptive | 16:9 for landscape and YouTube, 9:16 for TikTok and Reels, 1:1 for feed posts. |
| Resolution | 480p, 720p, 1080p, up to 2K | Draft on 480p or 720p, deliver on 1080p or 2K. |
| Duration (2.0) | 3s, 5s, 8s, 10s | Short, single-move clips and quick b-roll beats. |
| Duration (2.5) | 2.5s up to 30s, single clip | Multi-shot stories with scene changes, no stitching. |
| Frame rate & file | 24 fps, MP4 | Standard cinematic output on every platform. |
Copy-paste example prompts
Each block below is a full, ready prompt that stacks the tables into the formula. Change the subject and setting, then paste into Seedance and set the controls to match the parentheses.
1. Cinematic character b-roll
A young barista in a linen apron steams milk behind a marble counter, steam curling into warm morning light. Cozy specialty coffee shop, plants on the windowsill. Medium close-up, slow dolly-in on her hands and the pitcher. Golden hour light through the front window, soft shadows, shallow depth of field, cinematic. Ambient: quiet café murmur, an espresso machine hiss. SFX: milk steaming, a spoon tapping ceramic. No subtitles, no on-screen text. (16:9, 1080p, 8s)Why it works: a tight shot plus a single push-in and warm light turns a simple action into a mood.
2. Vertical product ad
A matte-black wireless earbud case sits on wet slate, lid slowly opening to reveal the earbuds glowing softly. Minimalist studio set with a single beam of light. Macro shot, slow orbit around the case. Low-key lighting with a hard rim light and volumetric haze, glossy reflections, premium and clean. Music: minimal, confident electronic pulse. SFX: a soft magnetic click as the lid opens. No subtitles, no on-screen text. (9:16, 2K, 5s)Best for: hero product shots for TikTok, Reels, and Shorts.
3. Talking character with lip-sync
A confident startup founder in a charcoal blazer speaks directly to camera in a bright modern office, glass walls behind her. Medium close-up, static locked-off camera. High-key three-point lighting, soft and even, crisp and professional. She says, "We built this because the old way just wasted everyone's time." Ambient: quiet office room tone. No subtitles, no on-screen text. (16:9, 1080p, 8s)Why it works: quotes drive automatic lip-sync, and a locked-off medium keeps focus on the delivery. See more in the best Seedance prompts roundup.
4. Multi-shot mini story
A hiker reaches a misty mountain ridge at dawn. Keep the same hiker and jacket across all shots. 0-3s: wide establishing drone shot pulling back over the ridge, blue hour light and low fog. 3-6s: medium tracking shot following behind the hiker as they climb, cool natural light. 6-10s: close-up on the hiker's face breaking into a smile as the sun crests, warm golden rim light. Ambient: wind, distant birdsong, crunching gravel. Music: uplifting cinematic strings that swell at the summit. No subtitles, no on-screen text. (16:9, 2K, 10s)Best for: a full beat, arc, and payoff in one Seedance 2.5 clip.
5. Reference-driven shot
A skateboarder rides through a sunlit downtown plaza at midday, weaving between planters. Use @Image1 as the skateboarder's appearance and match the color grade and film-grain style of @Image2. Copy the camera motion from @Video1. Full shot, tracking alongside the board. Bright natural light, soft shadows, energetic and candid. Ambient: city hum, distant chatter. SFX: wheels rolling over pavement, a sharp kickflip snap. Sync the cut to the beat of @Audio1. No subtitles, no on-screen text. (9:16, 1080p, 8s)Why it works: four references lock a face, a look, a motion, and a rhythm so the clip matches your brand.
6. Macro food close-up
Thick chocolate sauce pours slowly over a stack of pancakes, sliding down the sides and pooling on the plate. Warm rustic kitchen table with soft window light behind. Extreme close-up, static shot holding on the drip. Backlit natural light with a warm rim, glossy and appetizing, shallow depth of field. SFX: a slow, thick pour and a gentle sizzle. Music: light, warm acoustic guitar. No subtitles, no on-screen text. (1:1, 1080p, 5s)Best for: square feed posts and food-brand content.
Frequently Asked Questions
What is the Seedance prompt formula?
Subject + Action + Environment + Camera move & framing + Lighting + Style, then optional Audio (dialogue in quotes, SFX, ambient), and a settings note like (16:9, 1080p, 8s). Naming each part in that order gives Seedance the least to guess.
How do I add sound to a Seedance prompt?
Seedance 2.0 and 2.5 generate audio natively in a single pass, so you describe it in words. Put spoken lines in quotes for automatic lip-sync, then add SFX:, Ambient:, and a one-line music mood. Use @Audio1 to sync to a supplied track and add "no subtitles, no on-screen text" to keep captions off.
Do I set aspect ratio and resolution inside the prompt?
No. Aspect ratio, resolution, and duration are app or API settings in Dreamina, fal.ai, and similar tools — there is no --ar flag like Midjourney. End the prompt with a plain reminder in parentheses, such as (9:16, 1080p, 5s), so you set the controls to match.
What do @Image1, @Image2, @Video1, and @Audio1 mean?
They are multimodal references. @Image1 supplies a character's face and appearance, @Image2 sets the style or aesthetic, @Video1 supplies camera-movement or motion to copy, and @Audio1 supplies a music, rhythm, or dialogue track. Write them into the sentence, for example: Use @Image1 as the character's appearance.
How long can a Seedance clip be?
Seedance 2.0 makes clips of 3, 5, 8, or 10 seconds. Seedance 2.5 (announced June 2026) generates a single continuous clip up to 30 seconds with scene changes and tempo shifts, so you can cover a small story without stitching.
How do I prompt a multi-shot Seedance clip?
Use time-coded segments in one prompt, such as 0-3s: wide establishing shot… 3-6s: medium… 6-10s: close-up. Keep the subject and environment consistent across segments and give each one its own camera move. For short clips under about 5 seconds, use a single camera move instead.