Seedance 2.5 Prompt Guide: Create Videos using 50+ Prompts (2026)

ByteDance officially launched Seedance 2.5 on July 31, 2026, and the model reads a prompt differently than anything in the Seedance line before it. A 30-second single-pass generation, timestamp-level editing, and reference counts up to 30 images, 10 video clips, and 10 audio clips per generation mean a prompt now has to work as a shot brief for an entire scene, not a caption for one moment.

This guide covers exactly how to structure a Seedance 2.5 prompt, walks through ByteDance's own official example prompts from the launch announcement, and gives you more than 50 ready-to-adapt prompts across the categories creators actually need. For the full feature breakdown and how to access the model, see Seedance 2.5 on ImagineArt.

What Changed About Prompting for Seedance 2.5

Seedance 2.5 builds on Seedance 2.0's unified audio-video architecture, so the fundamentals of a good Seedance prompt still apply: a clear subject, concrete action, and named camera movement. What's new is scale and precision.

• 30 seconds per generation, with multi-round extension. A prompt can now describe an actual multi-beat story, setup, development, a turning point, resolution, rather than a single moment stretched thin. ByteDance's own materials describe this as organizing "multiple logically connected shots" within one generation.
• Reference inputs jump to 30 images, 10 video clips, and 10 audio clips. That's a genuine step up from Seedance 2.0's 9 images, 3 videos, 3 audio clips, and it means a prompt can now name a full cast, an environment, a motion reference, and a voice sample all in the same request, addressed directly by number (@Image 1, @Video 1, @Audio 1).
• Timestamp-level editing. You can now control narrative, camera perspective, and pacing for a specific time window inside a single generation, and edit a specific clip's characters, actions, or plot elements afterward without touching the rest.
• Clay render referencing. A textureless 3D blockout can now define camera movement, blocking, and spatial layout, letting the model treat your previz as a literal staging reference rather than something to invent from a text description.

Writing a Seedance 2.5 Prompt:

Every Seedance 2.5 prompt breaks down into the same core formula, and you can leave out any piece you don't need:

Subject + Action or Event + Scene and Environment (optional) + Visual Style (optional) + Camera Movement/Cut (optional) + Audio (optional)

The first part, subject plus action, is the only non-negotiable piece. Everything else sharpens the result but isn't required to get something usable.

• Subject + action or event states who or what is doing what. This is the foundation every other clause builds on.
• Scene and environment covers location, time, weather, spatial relationships, and background state.
• Visual style covers lighting, color, materials, image texture, and overall mood.
• Camera movement/cut covers shot size, camera angle, movement, focus subject, and any transitions.
• Audio covers dialogue, voice characteristics, ambience, sound effects, and music.

A basic prompt built on this template reads like a short paragraph, not a list of tags:

A ceramic artist finishes a pale blue cup in a studio at dawn, lifts it from the wheel, and places it in the center of a wooden shelf. Soft morning light enters through the window. The wet clay has a delicate sheen, and the workbench remains tidy. Begin with a medium shot of the wheel-throwing process, slowly push in toward the cup's surface texture, then cut to a frontal view of the shelf. Retain the low hum of the pottery wheel, the friction of clay, and subtle indoor ambience.

Generation parameters, resolution, duration, aspect ratio, don't belong in the prompt text itself. Those get set on the generation page or through the API separately.

Reference Materials: Limits and How to Actually Use Them

Seedance 2.5 combines up to 50 reference materials in a single generation, but that ceiling breaks down by type, and each type has its own recommended range for reliable output.

The recommended ranges exist for stability, not because the model refuses anything beyond them. You can push further: 9 to 12 subjects in subject images, 6 to 10 subjects in subject audio or video, 6 to 8 reference images for a video edit, but consistency tends to degrade as the material count climbs. If more than five subjects also need multiple views each, put each view in its own separate image rather than combining several views into one collage. Independent view images hold up more reliably than a collage does.

Defining what each reference actually contributes

This is the piece most prompts skip entirely, and it's the difference between a reference that helps and one that quietly contaminates the output. After uploading reference materials, state explicitly what each one contributes, and just as importantly, what to exclude. If a reference image's background, an unrelated person in frame, or an unwanted composition choice could bleed into the result, say so directly. The model won't reliably infer which person, prop, or scene a given material represents just from a label inside the image itself. That mapping has to live in the prompt text.

The template looks like this:

@Image 1 defines 's <appearance, clothing, structure, or material>.

@Video 1 defines <motion, camera movement, or pacing>.

@Audio 1 defines 's <voice, dialogue, ambience, or music>.

completes in . The visuals feature , with .

Applied to the ceramic artist example:

@Image 1 defines the ceramic artist's facial features, hairstyle, and dark green apron. Do not use the image background.

@Image 2 defines the wooden workbench, window placement, and morning light of the pottery studio. Do not use the people in the image.

@Video 1 defines the pacing of throwing clay with both hands, lifting the cup, and placing it down. Do not use the person's identity, clothing, or scene from the video.

The ceramic artist finishes a pale blue cup in the pottery studio at dawn, lifts it from the wheel, and places it in the center of a wooden shelf. Begin with a medium shot of the wheel-throwing process, then slowly push in toward the cup's surface texture. Retain the sound of the wheel, the friction of clay, and indoor ambience.

Notice the two exclusions doing real work here: without them, Image 1's background and Image 2's people could both end up in the final generation uninvited.

Multiple views of the same subject

If several images show different angles of the same person or product, say so explicitly rather than letting the model guess they're related:

@Image 1 defines the front view of the same folding desk lamp.

@Image 2 defines the left-side structure of the same folding desk lamp.

@Image 3 defines the right-side structure of the same folding desk lamp.

@Image 4 defines the rear structure of the same folding desk lamp.

All four images define one folding desk lamp. The output must contain only one lamp throughout.

Without that last line, there's a real risk the model treats four angles as four separate lamps instead of one.

When a reference video already defines the motion

If a reference video accurately captures the motion, camera movement, and sequence you want, state only which attributes to inherit from it rather than re-describing the action in full. Repeating the motion description in text can actually conflict with what the reference itself shows. That said, a blockout or motion-reference video mainly supplies motion and spatial structure; it doesn't replace defining the subject, scene, action, and visual style elsewhere in the prompt. The video handles movement; the text still has to handle everything else.

Special Syntax for Audio and Text

Prompts read as natural language by default, but when you need to separate music, sound effects, dialogue, and on-screen subtitles unambiguously, four bracket types do that job:

Mixing plain descriptive language with these markers in the same prompt is fine; the brackets exist specifically for the moments where ambiguity would otherwise cost you a regeneration, not as a requirement for every clause.

Reinforcing dialogue language and accent

When dialogue isn't in Chinese, name the language directly before the line:

The girl says softly in Japanese: {もう大丈夫です}

If English dialogue text keeps coming out spoken in Chinese, or you need a specific regional accent rather than a generic delivery, reinforce it using this structure:

Dialogue language + regional variety or accent + delivery style + speaker + {dialogue}

Dialogue language: American English. The girl says in natural, conversational American English: {I thought you weren't coming.}

Dialogue language: authentic Los Angeles English. The young man says in natural Los Angeles vernacular: {No way, you actually made it.}

How to Structure a Seedance 2.5 Prompt

A working framework for organizing a Seedance 2.5 prompt breaks into five parts:

Weak vs. Strong Seedance 2.5 Prompts

The difference between a prompt that produces something usable and one that doesn't usually comes down to specificity in exactly these five areas.

Common mistakes to avoid

• Describing a still image instead of motion over time. A video model needs to know what changes across the full duration, not just what a single frame looks like.
• Vague camera direction. "Cinematic" tells the model nothing actionable. Name the actual shot and the specific move.
• Cramming an entire multi-scene story into one prompt without timestamps. Use the timestamp-level editing structure to assign specific beats to specific time windows instead.
• Leaving reference assets unlabeled. If you supply five images, say explicitly what each one is for: character, location, palette, or product, rather than letting the model guess which reference drives which part of the scene.

Official Seedance 2.5 Example Prompts (From ByteDance's Launch Announcement)

These prompts come directly from ByteDance's official Seed team blog post announcing Seedance 2.5, and they demonstrate the model's actual new capabilities rather than a third party's guess at what it can do.

1. One-take backstage-to-stage tracking shot (T2V)

One-take handheld gimbal tracking shot. The camera slowly pushes in through a gap in a heavy red curtain and enters a warm-toned backstage dressing room. A young female singer, with her back to the camera, is adjusting her earpiece as a staff member reminds her it's time to go on. She turns toward the camera and starts singing citypop. The camera pulls back and tracks her as she passes through the curtain into a dim backstage corridor, interacting naturally with her dancers along the way; one staff member hands her a microphone. She and the dancers then step onto the stage, and the camera arcs around to the back, gradually revealing the red-and-black stage design, LED screens, spotlights, haze, and reflective floor. The camera finally pulls out to a wide shot of the arena, showing the packed audience, light boards, glow sticks, and cheering crowd, capturing the youthful, free-spirited climax of the concert.

▶ Watch the example video

2. Multi-round extension for a chase sequence (R2V)

Extend the video. Continue from the visuals and subjects in @Video 1 and generate another 30-second clip, keeping the character subjects, scene, visual style, and sound effects consistent. The little boy runs along the train carriage holding a soccer ball. When the subway stops, the side door opens and he immediately dashes out, with the male lead chasing after him. The two run across the platform and out onto the street, startling passersby and vehicles along the way. The male lead finally catches up and grabs him. The boy looks up, aggrieved. The male lead's anger slowly fades; he pats the boy's head and shows a helpless smile.

3. Multi-character concert scene with 18 references (R2V)

A 30-second concert sequence in 16:9 landscape, with cinematic realism, authentic concert hall lighting and shadows, warm golden stage lighting, and the atmosphere of a formal classical concert. Use @Image 1 for the venue. Reference @Image 2 for the pianist. Reference @Image 3 for the cello. Reference @Image 4 for the violin. The lead vocalist must strictly follow @Image 5. Reference @Images 6 to 10 for the rest of the orchestra. Reference @Images 11 to 14 for the choir. Reference @Images 15 to 18 for the audience seating. The lead vocalist walks from center stage toward the front edge. The pianist is positioned by the piano. The orchestra is arranged on both sides and toward the rear. The choir stands at the back of the stage. Open with a high-angle wide shot of the full concert hall. The pianist strikes the keys, and the lead vocalist steps into the spotlight and begins singing. The camera naturally moves across the violin, cello, and orchestra as they perform together, with the violin feeling bright and the cello warm. In the latter part, the choir joins in.

https://blogs-cdn.imagine.art/Multi_character_concert_scene_with_18_references_R2_V_dac2e9b789.mp4

4. Clay render reference for a fantasy sequence (R2V)

Refer to @Clay Render 1 for camera movement, pacing, shot-size transitions, subject trajectory, and blocking. Refer to @Image 2 for character design, scene, materials, lighting, color, and fairy-tale atmosphere, and render the white model as a dreamy, warm, 3D animated short with a childlike fantasy feel. The story unfolds as follows: flight through a fantasy sky, mythical beasts flying alongside through a sea of clouds, a dive into the ocean, weaving through the deep with manta rays, passing through a mirrored rift in spacetime, picking stars from the cosmos, transforming back into the bedroom, father tucking in the blanket, the picture book closes and holds on the final frame.

5. Green screen background replacement with timestamps (R2V)

Using @Video 1, render the green-screen background, obstacles, wardrobe, and supporting characters. 0-4s: outdoor training, replace the obstacles with rocks, bricks, tires, and wooden crates. 4-10s: locker room, friends offering encouragement. 10-15s: international match, replace the training poles with original defenders and a goalkeeper, and the protagonist scores. Overall photorealistic, cinematic quality.

6. Camera-only edit on an existing clip (R2V)

Edit @Video 1. Keep the characters, actions, and visual style unchanged. Adjust only the camera movement. A 15-second segmented camera plan: 0-4s, a micro-FPV move skims tightly past the pan, then follows the popping toast and whip-pans to the coffee; 4-7s, push in and track laterally along the rim of the pan, following the fried egg as it flips up and lands back in place; 7-11s, rapidly rise to a top-down view, then descend at a steady pace, sweeping across the plate and keys; 11-15s, use a handheld close-up to follow the hands with a fast lateral whip, then push in on the breakfast and pull back to a medium two-shot.

7. Clay render for industrial or product sequences (R2V)

Reference the camera work, composition, shot scale, spatial relationships, part positions, model structure, assembly order, and motion paths from @Clay Render 1. Reference the materials, lighting, color, reflections, and atmosphere from @Image 1, and turn the clay render into a high-end, photorealistic car assembly sequence.

8. Historical or educational narrative (R2V)

Expressive Eastern painterly style. A street scene in Lin'an during the Southern Song dynasty. Several children run and shout through the bustling street. The camera follows the children as they run, sweeping past the lively street. The camera then tilts up to reveal a scholar figure from @Image 1. He turns his head, and in the distance stands a man among the fading lantern lights. The shot stays continuous throughout.

https://blogs-cdn.imagine.art/Historical_or_educational_narrative_R2_V_98dc9df0b1.mp4

9. Peking Opera single continuous take with three character references (R2V)

16:9 widescreen, cinematic texture, single continuous take, smooth camera movement, no cuts. Scene reference: @Image 4. 0-5s: Open with a close-up of the lead performer from @Image 2. The camera slowly circles his upper body and transitions into a medium shot, his robes sweeping past the lens to form a natural occlusion, and the camera follows through to the second performer's side in @Image 1. 6-10s: The camera steadily circles the second performer in a medium shot from @Image 1, following her water sleeves through the arc. 11-20s: A third performer from @Image 3 enters with an aerial flip. At the end, all three face the audience and strike a synchronized finale pose.

10 Cinematic Seedance 2.5 Prompts for Common Use Cases

These prompts follow the same directorial structure, subject, action over time, camera and lighting, synchronized sound, adapted for the scenes creators most often need. They work on Seedance 2.0 as well as 2.5, so you can test the timing and reference structure before generating your final take.

10. Baby video (emotional character drama)

STYLE: Anamorphic 35mm, photorealistic, warm commercial grade, fine grain. Toys are real physical toys, baby is real — scale achieved through forced perspective and ultra-low camera placement making the baby enormous relative to the toy world.

GRADE: Warm indoor afternoon — soft window light from one side, warm amber, rich saturated toy colours — primary reds blues yellows popping against warm wooden floor. Skin tones golden and warm.

CAMERA: Ultra-low toy-level throughout the chaos — lens at toy height making everything enormous. Fast handheld during chaos, locked wide on scale shots, slow warm dolly on the final scene.

SUBJECT: A colourful toy world — toy cars, lego bricks, toy people figures, small lego buildings on warm wooden floor. A baby, approximately 10 months, white onesie, crawling.

00:00-00:03 > Ground level — the toy city fills the frame enormous, toy cars at a tiny intersection, lego buildings like skyscrapers, toy people going about their day. Peaceful. Then the floor SHAKES. A toy car tips. A lego brick falls. The toy people wobble.

00:03-00:06 > Bigger shake — a lego building collapses in slow motion, bricks scattering. Toy cars crash. Camera at ground level tracking the chaos — a toy car spinning past the lens, lego bricks flying overhead, a toy figure falling face-first in the foreground. BOOM. BOOM. Each impact bigger. Remaining buildings crumble. Lego pieces raining. Toy people running in every direction at ground level. The massive shadow falls across the toy world. Everything goes quiet.

00:06-00:10 > Camera tilts up slowly from toy level — past scattered lego, past crashed cars, up and up — and there, enormous against the warm window light, the baby. Backlit, soft and golden, crawling toward the toy world. Each hand coming down — BOOM — lego bricks jumping on impact. From this angle the baby is the size of a building. Face completely delighted, eyes wide, drooling slightly, moving with total unstoppable purpose.

00:10-00:13 > The baby's massive hand descends from above in slow motion, fingers closing around a toy man figure. Camera at toy level watching the toy man rise upward, getting smaller as the baby lifts it. Cut to the remaining toy people watching from ground level — frozen plastic faces directed upward in absolute awe. The baby examines the toy man, turns it, studies it. Then — directly into the mouth.

00:13-00:16 > CUT — completely different energy. Baby sitting on the wooden floor, calm and happy, surrounded by scattered toys. Warm afternoon light falling soft. He bumps a toy car gently into the toy man — brrm brrm — then picks the toy man up and inspects it again. Fully absorbed. Peaceful. The chaos is over.

00:16-00:19 > Camera slowly widens — warm room, afternoon light, the baby small in the frame surrounded by scattered coloured toys on the wooden floor. A slow dolly push-in from the front — the baby looks up directly at camera, makes eye contact, gives one wide gummy smile. Toy man still in his fist.

00:19-00:22 > Clean warm hold. Tagline appears in simple warm typography: Big imaginations need bigger pieces. Fade warm to white.

AUDIO: BGM — opens silent, low rumbling bass on the first shake, builds into fun dramatic orchestral — big brass and percussion playing the disaster completely straight, baby's arrival announced by deep timpani. Transitions to warm gentle playful piano and soft strings as the baby sits and plays. Near-silence on the tagline. SFX — toy cars crashing, lego bricks scattering on wooden floor with satisfying plastic clatter, deep physical BOOM on each baby crawl step, baby giggle on the smile, toy car brrm on the playing section. All SFX at the scale of the toy world — the baby's sounds enormous.

11. Victorian house in night fog (horror or thriller teaser)

A pale pink Victorian house stands in dense night fog, glowing with warm light from curtained windows. Slow dolly forward along the front path, volumetric fog catching the glow, cool moonlit rim light playing against the warm interiors. Distant wind chimes and muffled night ambience rise gently as the camera approaches the porch.

https://blogs-cdn.imagine.art/Arooj_Ishtiaq_A_pale_pink_Victorian_house_stands_in_dense_night_fog_glowing_wi_87aa69e635.mp4

12. Neon city sleeper (calm cyberpunk world-building)

A person sleeps in a bed facing a large window overlooking a vibrant neon-lit cityscape at night. The camera drifts slowly from over the bed toward the glass, city lights pulsing softly across the sheets. Rain streaks the window; distant traffic and a low synth hum give the scene a calm, cyberpunk mood.

13. Retro headlights ignition (automotive brand drama)

Extreme close-up of dual square headlights emitting blinding white light on a red and grey retro vehicle. The lights flare on with a mechanical clunk, lens flares blooming across chrome trim. Slow pull-back reveals wet asphalt reflections while an engine rumble builds beneath an electronic ignition chime.

14. Floating product reveal (ecommerce launch)

Floating modern sneaker with a textured upper and a translucent sole plate with traction studs, clean white studio background. The product rotates slowly in zero gravity while macro details resolve: stitching, texture grain, light refracting through the sole. Orbiting camera ends in a snap to a centered hero frame, punctuated by a clean whoosh and a click on the lock-in.

https://blogs-cdn.imagine.art/Arooj_Ishtiaq_Floating_modern_sneaker_with_a_textured_upper_and_a_translucent_s_8149ee2f11.mp4

15. Luxury watch macro story (premium detail advertising)

Extreme macro shot of a luxury mechanical watch on a dark velvet surface. The second hand ticks with satisfying precision as the camera slowly orbits 360 degrees, revealing intricate engravings and sapphire crystal reflections under soft studio key lighting. Tiny gears visibly turn inside the open caseback. Delicate ticking sound builds into a resonant chime at the hero frame. Cinematic shallow depth of field, premium jewelry photography style.

16. Fashion Film

STYLE: Sony DCR-TRX handycam 2003 — SD resolution, CCD bloom on highlights, autofocus hunting, digital zoom artifacts, slight interlace feel, date-time stamp bottom right in small green font. Handheld unsteady, candid friend-on-the-street energy.

GRADE: Mid-noon San Francisco hazy sunshine — blown-out warm whites in sky, overexposed pavement, skin tones warm and slightly oversaturated in the handycam way, soft warm shadows, everything luminous and washed. His deep red outfit cutting vivid through the haze.

CAMERA: Handycam throughout — digital zoom, autofocus hunting, natural shake, occasional rack focus. Kinetic fast cuts mid-section. No stabilisation. Real street feel.

SUBJECT: Black male, powerfully built — deep red cropped leather moto jacket, white shirt, red NY Yankees tie, cream distressed cargo pants with chain hardware, red patent duffle bag, red beret, tinted sunglasses, chunky silver jewellery, phone to ear, white chunky sneakers. SF downtown, mid-noon.

00:00-00:03 > Camera finds him from across the street — digital zoom hunting before locking on him mid-walk, phone to ear, red duffle swinging, traffic occasionally cutting in front of the shot. He has no idea he's being filmed. Noon light washing overexposed behind him.

00:03-00:05 > Operator crosses and closes in — tracking behind and to the side at chest height, camera bobbing with walking pace, red jacket and beret vivid in the haze. Autofocus briefly loses him on a passing pedestrian then snaps back.

00:05-00:07 > CUT — shooting through a gap between parked cars, low angle, his legs and chain hardware and sneakers in the lower frame, upper body above, SF street behind him.

00:07-00:09 > CUT — tight digital zoom on just the sneakers and cargo chain details catching the noon sun. CUT — zoom pulling wide fast to show his full figure from behind, duffle swinging.

00:09-00:10 > CUT — side angle tight on his face above the sunglasses, jaw moving on the phone, beret sharp against the blown-out sky.

00:10-00:15 > KINETIC SECTION — he stops at a corner, takes the phone from his ear, looks left then right. Fast cuts all within this moment: left angle waist-up — sunglasses tight — low looking up full figure — NY tie and chain close — hand adjusting the duffle — face looking right — low wide — red jacket collar tight. Each cut under one second, handycam whipping between angles with motion blur, noon light consistent and hot. Ends wide low angle looking up as he starts walking again.

00:15-00:19 > Long tracking shot from behind — camera at shoulder height, the SF street opening ahead, his shadow stretching forward on the hot pavement, the red outfit vivid and moving, other pedestrians and real street life around him. Unhurried.

00:19-00:22 > CUT — side tracking from a distance, SF architecture behind him, hazy light washing the building faces bright. CUT — tight on the chain hardware at his waist swinging with each step, noon light catching each link.

00:22-00:25 > CUT — tight on his face, autofocus finally perfectly locked, sunglasses catching the bright noon sky as a double reflection. He still doesn't look at the camera.

00:25-00:27 > He walks toward and past the camera — the red jacket filling the entire frame for one frame as he passes — then the camera turns following him from behind as he walks away.

00:27-00:30 > Camera pans off him to the right — drifting along the building wall beside him, slightly unsteady, noon light warm on the concrete — until the pan finds: LIVON engraved large and raw into the wall. Autofocus locks on the letters. Date-time stamp in the corner. Hold. Tape cuts to static.

AUDIO: BGM — early 2000s pop track, bright and slightly compressed, melodic with light percussion and warm synth, playing as if from a nearby portable speaker recorded through the handycam mic — slightly tinny and real. Fades on the wall hold. SFX — SF street ambient throughout, passing cars, distant voices, handycam auto-zoom whir on the first shot, autofocus click hunting and locking, chain hardware faint jingle on kinetic cuts, tape static cut at the end.

17. Cyberpunk street chase (fast-paced action)

A hooded figure sprints through a rain-slicked neon megacity alley at night, holographic billboards reflecting in puddles. Handheld tracking shot follows from behind, then pushes forward as the runner vaults over a hovering drone. Blue and magenta neon rim lighting, lens flares on wet surfaces, heavy breathing and distant siren wail synced to footsteps and impacts.

18. Fantasy portal awakening (magical story reveals)

A curious young girl in a flowing cloak stands in an ancient forest glade at twilight. As she reaches toward a glowing rune stone, a radiant portal swirls open with swirling golden particles. Slow push-in combined with gentle upward crane shot reveals magical creatures emerging. Ethereal wind chimes and soft orchestral swell synchronized with the light burst. Warm volumetric god rays, dreamy bokeh, magical realism, 35mm cinematic look.

https://blogs-cdn.imagine.art/Arooj_Ishtiaq_A_curious_young_girl_in_a_flowing_cloak_stands_in_an_ancient_fore_3aca229405.mp4

19. Desert dusk caravan epic (historical and travel-scale drama)

A majestic camel caravan winds through vast golden sand dunes at sunset. The camera starts with a sweeping aerial dolly revealing the full line of travelers, then slowly descends to track alongside the lead rider on a white camel. Billowing robes catch the wind, dramatic rim lighting from the setting sun. Distant camel bells and low desert wind rise in volume as the caravan crests the final dune. Cinematic anamorphic lens flares, rich color grading, subtle film grain.

Prompts for Product and Ecommerce Video

20. A skincare bottle rotates on a marble surface under soft window light, condensation beading on the glass. Camera performs a slow 360-degree orbit at eye level, ending on a straight-on hero frame. Ambient room tone with a soft chime on the final lock-in.

21. A leather handbag sits open on a linen backdrop as a hand reaches in to reveal the interior lining. Macro push-in on the stitching detail, then pull back to a three-quarter hero angle. Soft studio ambience, a subtle fabric rustle synced to the reveal.

22. A wireless earbud case opens in slow motion, the lid lifting to reveal the buds nestled inside with a soft LED glow. Camera tilts down from a high angle to a level macro shot. A clean mechanical click on the lid opening, synthesized brand chime on the reveal.

23. A bottle of perfume is lifted from a bed of rose petals, light catching the faceted glass as it rises. Slow vertical camera rise following the bottle, ending on a centered product shot with soft bokeh petals in the foreground. Delicate string swell timed to the lift.

24. A running shoe splashes through a puddle in slow motion, water droplets suspended mid-frame. Side-tracking camera at ground level, transitioning to an orbit around the suspended droplet moment. Sharp splash sound stretched into the slow-motion beat, then a clean product-reveal chime.

25. A coffee bag is torn open, beans cascading into a glass container below. Overhead camera angle transitioning to a side view tracking the falling beans. Rich cascading sound design, ending on a warm ambient café tone.

Prompts for UGC-Style and Social Ad Video

26. A woman in a bright kitchen holds up a skincare product to camera, speaking directly into the lens with an enthusiastic but natural tone about how it changed her morning routine. Handheld selfie-style camera, slight natural shake, warm morning light through a nearby window. Casual room tone, clear spoken dialogue.

27. A young entrepreneur sits in a cafe explaining a mobile app feature to camera, gesturing naturally as he talks. Static medium shot at eye level, shallow depth of field with the cafe blurred behind him. Ambient cafe chatter under his voice, natural conversational pacing.

!image.png

28. A fitness creator demonstrates a resistance band exercise in a home gym, explaining the movement between reps. Handheld tracking shot following the movement, natural gym lighting. Breathing and effort sounds synced to the exercise, brief spoken explanation between sets.

29. A food creator takes a first bite of a new snack product, reacting with genuine surprise before describing the flavor to camera. Close-up handheld shot, kitchen counter setting with warm practical lighting. Crunch sound synced to the bite, natural spoken reaction.

!image.png

30. A traveler unboxes a packing cube organizer on a hotel bed, showing how it fits into a suitcase. Overhead handheld angle transitioning to a side view of the packing demonstration. Fabric rustle and zipper sounds, casual narration.

Prompts for Fashion and Beauty Video

31. A model walks toward the camera down an empty urban street at dusk, an oversized coat billowing with each stride. Steady tracking shot at hip height, moving backward at the model's pace. City ambience, footsteps synced to the stride, coat fabric movement audible.

32. A hand applies liquid foundation to a cheek in extreme macro, the product blending seamlessly into the skin. Static macro shot, soft diffused beauty lighting. Subtle product application sound, calm ambient studio tone.

33. A model turns in a circle in a flowing evening gown under a single spotlight, fabric lifting with the rotation. Camera orbits slowly around the model at the same pace as the turn. Fabric movement audible, a low orchestral swell building through the rotation.

!image.png

34. Animated Lady Video

STYLE: Two worlds — photorealistic opening transitioning into flat bold 2D illustration matching the reference exactly: thick black outlines, flat colour fills, crayon-textured strokes, white background world.

GRADE: Real world — warm European afternoon, golden light, cobblestone. Illustration world — pure white background, flat vivid primaries matching the reference: pink skin, orange curly hair, blue wide-leg jeans, red chunky boots, confetti-dot white jacket.

CAMERA: Warm handheld tracking in the real world, smooth fluid animation camera in the illustration world — push-ins, playful pans, following her energy.

SUBJECT: European girl, mid-20s, blonde, casual streetwear, German cobblestone pathway. A bright gummy pack. Her illustrated alter-ego — the reference character exactly.

00:00-00:03 > Warm handheld tracking — she walks along a sunny German cobblestone path, golden afternoon light, European architecture behind her. She reaches into her bag, pulls out the gummy pack, pops one in. Chews once. Eyes close. A soft smile. She looks at her hand. Something is different.

00:03-00:07 > THE TRANSITION — slow and magical. The cobblestones around her feet begin illustrating themselves — flattening into thick drawn lines, buildings behind her simplifying into bold outlined shapes, colour bleaching out to white. Her skin warms to illustrated pink, clothes flatten into vivid flat fills, orange curly hair pops into place, rosy cheeks appear, green earrings lock in. The world is now white. She is now the character. Fully. She looks at her illustrated hands, wiggles her fingers, looks at camera and grins wide.

00:07-00:11 > A beat drops and she MOVES — dancing in the white space, confetti dots scattering off her jacket as she spins, red boots stomping flat satisfying impacts on the white ground, orange curls bouncing. The illustrated world builds around her — round chunky thick-outlined dogs trotting past, a tiny illustrated car puttering by and beeping, flowers pushing up from the white ground where she steps. Everything bold and flat and alive in the same style.

00:11-00:14 > She crouches and pets one of the round dogs — it spins in a happy circle. She laughs — illustrated laugh lines bursting from her face. She pulls another gummy mid-dance, holds it up — a tiny flavour burst radiates from it in the illustration style. She pops it in. Eyes close in illustrated bliss, little stars floating up from her head.

00:14-00:16 > Wide white frame — her full illustrated figure centre, the dogs and car and flowers around her, joyful and vivid. She does one final spin, confetti dots raining across the frame. Then she stops. Points directly at camera — the reference pose exactly.

00:16-00:18 > The gummy pack floats into frame beside her in the same illustration style, product label vivid and readable. Tagline appears in bold illustrated hand-lettered font: INDULGE IN FUN. EXPLORE EVERY FLAVOUR. Brand logo beneath it. Hold. Fade to white.

AUDIO: BGM — warm street ambient at open, single playful piano note as she eats the gummy, then a bright bubbly pop track detonates on the illustration transition — upbeat, fun, melodic, energetic throughout. Drops to a playful minimal sting on the tagline. SFX — cobblestone footsteps, gummy packet rustle, soft magical shimmer through the transition, illustrated boot stomps with flat cartoon thuds, round cute dog bark, tiny car beep, flavour burst shimmer on the second gummy, confetti shower on the spin, freeze-frame pop on the final pose.

Prompts for Food and Culinary Video

35. A chef plates a dessert in extreme slow motion, chocolate sauce drizzling in a precise pattern across the plate. Overhead macro shot, warm kitchen lighting. Drizzle sound stretched into the slow motion, ending on ambient kitchen tone.

36. Steam rises from a bowl of ramen as chopsticks lift noodles into frame. Close side angle at table height, shallow depth of field on the steam. Ambient restaurant chatter, a soft slurp synced to the lift.

37. A pizza is pulled from a wood-fired oven, cheese still bubbling, and set down on a wooden board. Camera tracks the pizza from oven to table in one continuous move. Crackling fire ambience, sizzling cheese sound, ending on a satisfied exhale.

Prompts for Nature, Travel, and Documentary Video

38. A humpback whale glides slowly past a free diver suspended in clear blue open water, sunlight filtering down in visible shafts. Wide static underwater shot, the diver small in frame against the whale's scale. Muffled underwater ambience, low whale song.

39. A hot air balloon drifts silently over a valley of vineyards at sunrise, mist still clinging to the rows below. Aerial tracking shot following the balloon's slow drift. Distant burner flame sound, soft wind, quiet morning ambience.

40. A herd of wild horses runs across an open plain at dusk, dust rising behind them in the low light. Low tracking shot at ground level moving with the herd. Hoofbeats building in volume, wind across the plain.

41. A lighthouse beam sweeps across a rocky coastline in dense fog, waves crashing below. Slow static wide shot holding on the lighthouse, the beam rotating through frame. Foghorn in the distance, waves against rocks, gulls calling.

Prompts for Sports and Action Video

42. A skateboarder lands a trick down a set of stairs in a sunlit skate park, wheels catching the landing cleanly. Low tracking shot following the approach, then a static hold on the landing. Wheels on concrete, a crowd cheer on the landing.

43. A swimmer dives off the starting block in slow motion, water exploding around the entry point. Side-on camera at water level, transitioning to underwater as the swimmer submerges. Splash sound stretched through the slow motion, muffled underwater tone on the cut.

44. A boxer throws a combination in a dimly lit gym, sweat visible under a single overhead light. Handheld close-tracking shot circling with the footwork. Glove impact sounds synced to each punch, heavy breathing between combinations.

45. A cyclist descends a mountain switchback road at speed, leaning into each turn. Aerial drone shot tracking from above, descending with the road. Wind rush building with speed, tires on asphalt through each corner.

Prompts for Fantasy and Sci-Fi Video

46. An astronaut walks through a vast, dimly lit hangar bay lined with dormant spacecraft, dust particles floating in shafts of light from above. Slow tracking shot moving alongside the astronaut, camera at hip height. Distant mechanical hums, footsteps echoing in the cavernous space.

47. A dragon lands on a castle rampart at night, wings folding as its scales catch torchlight. Low-angle static shot holding on the landing point, camera tilting up as the wings fold. Wind gust from the landing, wing-fold sound, distant torch crackle.

48. A spaceship's engines ignite in a dark hangar, light flooding the space as it begins to lift. Wide static shot holding on the ship, camera slowly pulling back as the light expands. Building engine hum, mechanical clunks, rising to a roar on liftoff.

49. A wizard raises a staff and a storm gathers overhead, lightning arcing between clouds. Low-angle shot from behind the wizard, camera slowly rising with the gathering storm. Wind building, distant thunder, a resonant magical hum synced to the staff's glow.

Prompts for Corporate and SaaS Video

50. A team gathers around a conference table as a presenter gestures toward a screen displaying a product roadmap. Static medium shot from the side of the room, natural office lighting. Ambient office tone, clear spoken presentation audio.

51. A laptop screen displays a dashboard interface as a cursor moves through several menus, each click registering a subtle highlight. Static overhead shot of the laptop on a clean desk. Soft keyboard and click sounds, calm ambient office tone.

52. A founder walks through a modern office space explaining company values directly to camera, employees working in soft focus behind them. Steady tracking shot moving alongside the founder. Ambient office chatter under clear spoken narration.

Prompts Using Region Editing and Timestamped Control

These follow the same structure as ByteDance's own green-screen and camera-edit examples above, adjusted for common production needs.

53. Edit @Video 1. Keep the character, action, and camera movement unchanged. Change only the product held in the subject's hand to the item shown in @Image 2, matching the original lighting and shadow direction.

54. Using @Video 1, replace only the background environment with the location shown in @Image 3, keeping the subject, their clothing, and their motion completely unchanged. Match the new environment's ambient lighting to the subject.

55. Edit @Video 1. 0-5s: keep unchanged. 5-10s: replace the on-screen signage text with the text shown in @Image 4, matching the original typography style and lighting. 10-15s: keep unchanged.

Conclusion

A strong Seedance 2.5 prompt reads like an actual shot brief:

• A concrete subject
• Motion described across the full duration rather than a single frame
• A named camera move
• Clearly labeled reference roles when you're using more than one asset.

The 50-plus prompts above cover the categories most creators actually need: product reveals, UGC-style ads, fashion, food, nature, action, fantasy, and corporate video, and the official examples from ByteDance's own launch announcement show exactly how far the model's new reference and editing capabilities can be pushed when a brief is structured correctly.

Start with a prompt close to your actual use case, adapt the subject and setting, and test on Seedance 2.5 directly rather than assuming a template will work unchanged for your specific brand or character. For getting the reference and editing structure right on your first attempt, how to use Seedance 2.0 covers the same underlying multimodal input system this guide builds on, and Seedance 2.5 vs Seedance 2.0 covers which version actually fits your specific workload.

Frequently Asked Questions

How many reference images can I use in a Seedance 2.5 prompt?

Seedance 2.5 accepts up to 30 images, 10 video clips, and 10 audio clips as reference materials in a single generation, a significant increase from Seedance 2.0's 9 images, 3 videos, and 3 audio clips. Label what each reference is for directly in the prompt: character, location, palette, or product, so the model applies each one correctly rather than guessing.

What's the maximum length of a single Seedance 2.5 generation?

Up to 30 seconds in one continuous pass, with support for multiple rounds of extension afterward that maintain consistency of characters, environment, and pacing, allowing multi-minute output without stitching separate clips together.

Do Seedance 2.5 prompts work in Seedance 2.0?

Many of them do, particularly single-scene cinematic prompts that don't rely on 2.5-specific features like clay render referencing or timestamp-level editing. Testing a prompt's timing and reference structure on 2.0 before moving to 2.5 is a reasonable way to validate a brief cheaply.

What is clay render referencing?

A textureless 3D blockout, the kind of white-model previz a director or art department might already build, can be supplied as a reference to define camera movement, blocking, subject trajectory, and spatial layout. The model then renders a fully textured, lit scene that follows that exact staging rather than inventing its own composition.

Can I edit part of a video without regenerating the whole thing?

Yes. Seedance 2.5 supports timestamp-level and region-level editing, letting you change a specific character, action, background, or on-screen element within a defined time window while the rest of the clip stays unchanged. This includes green screen background replacement and camera-movement-only edits on existing footage.

Where can I use Seedance 2.5?

Seedance 2.5 is available on ImagineArt alongside Seedance 2.0 and Seedance 2.0 Mini for faster, lower-cost generation. For the full feature comparison between the two Seedance versions, see Seedance 2.5 vs Seedance 2.0.