Seedance 2.5 video prompts, each shown next to the clip it produced and credited to its author. Where the creator published their prompt we keep it verbatim; where they did not, we reverse-engineer one from the finished clip and label it as such. Hover to play, one click to copy.
Seedance 2.5 is ByteDance's video model, unveiled on 23 June 2026 at the Volcano Engine FORCE conference. On the shipped API it generates 4 to 30 seconds in a single pass — or duration -1 to let the model choose the length — and accepts up to 50 reference assets in one job: up to 30 images at 4K each, up to 10 video clips and up to 10 audio clips totalling 30 seconds each. It adds video editing and video extension as task types, and generates audio in the same latent space as the picture. Resolution is 480p or 720p; the native 4K at 10-bit shown at the FORCE launch is not exposed on the API. ByteDance also published an official prompting guide for it. This page collects Seedance 2.5 video prompts with the clip each one produced, free to copy. Since 8 August 2026 you can run Seedance 2.5 on apimodels.app as the model seedance-2.5, from $0.134/s; the creators in this library ran these prompts in Dreamina (Jimeng) and Higgsfield.
Source: ByteDance Seed (the model vendor’s official site)
Hover to preview · click for sound
The cat that has changed pose every time she walks past
Reconstructed
A girl carries a stack of books along the wooden corridor of an old Japanese house on a summer afternoon, passing the same sleeping cat four times; the cat is in a completely different sleeping position each time, and she never breaks stride until the last pass.
<GIRL> is a child of about ten in a white short-sleeved blouse and a blue jumper skirt, barefoot, hair tied back. <CAT> is a plump grey-and-white shorthair asleep on a blue floor cushion in the middle of the corridor. The house has shoji screens, tatami, an old electric fan turning in the background, and an open side that looks onto a bright green garden.
Pass one: <CAT> is curled into a tight circle, tail over nose. Pass two: <CAT> is stretched out flat on one side, front legs extended. Pass three: <CAT> is fully on its back, all four paws splayed upward, belly exposed, the tip of its tongue showing. Pass four: <CAT> has not moved at all and <GIRL> finally stops, kneels, and buries her face in its belly, smiling with her eyes shut.
The image is hand-painted 2D anime in a warm summer palette: visible brush texture in the greenery, soft cel shading, dust motes in low afternoon light, a faint rainbow flare where the sun crosses the lens, and gentle bloom on the paper screens.
The camera holds a locked-off wide of the corridor for each pass so the only thing that changes is the cat, cutting to a slow push-in on <CAT> after each pass to show the new pose, then finishing on a close-up of <GIRL>'s face against fur.
Keep the corridor layout, cushion position, fan and garden identical in all four passes. Keep <CAT>'s markings — grey saddle and cap, white chest, white paws — the same in every pose.
Sound includes cicadas, the fan's slow oscillation, bare feet on old wood, the paper screens rattling faintly, one small sleepy chirrup from <CAT>, and a sparse piano figure with long gaps. No dialogue.
Hand-drawn, fully fluid transitions with no mechanical cuts
Core Style: Fast, vivid, unstable. Fully hand-drawn with rich organic textures. Every transition is fluid and natural—never rigid or mechanically cut. The entire frame, including the background, remains in constant motion, driven simultaneously by emotion, gravity, and heat. Every frame should feel alive, unpredictable, and physically reactive.
[0–7.5s: Origin and Chain Reaction]
[0–1.5s]
A central abstract figure erupts into violent motion. Its emotional color core radiates outward as irregular pulsating waves while its silhouette tears apart in multiple directions. The background bends around the figure like compressed soft paper. Fine hand-drawn textures crawl rapidly across the entire frame, moving significantly faster than the figure itself, creating intense tension between body, skin, and surface.
[1.5–3s]
The separated contours accelerate into nervous linear forms. The figure's colorful core stretches into sharp ribbons and liquid stains. Some ribbons snap backward while others whip violently forward, leaving trembling trails behind. Empty spaces between movements ignite into new forms that emerge to fill emotional voids.
[3–4.5s]
Every shape surges diagonally across the composition. One massive form folds like fabric, revealing layered textures of grain, chalk, wet ink, scratched paint, and rough pigment beneath its surface. The background folds together with it, twisting into entirely new color fields. Smaller forms are dragged into constantly evolving speed lines.
[4.5–6s]
The image fractures into four emotional forces: calm, violence, vulnerability, and ecstasy. Each absorbs color from the others. Shadows appear before their objects, behaving like prophecies. The original forms chase these shadows, but the shadows mutate faster—becoming brighter, thicker, and more dominant until they surpass the objects themselves.
[6–7.5s]
A rapid chain reaction begins. Colliding forms generate visual motifs derived from the reference imagery: fragments, symbols, silhouettes, stains, and expressive linework. These newly born motifs accelerate in different directions. Frozen traces solidify into new entities before melting again and splitting into tiny emotional echoes.
---
[7.5–15s: Compression and Extreme Transformation]
[7.5–9s]
The entire frame collapses toward an invisible point beneath the center. Every visual layer compresses inward at a different speed. Maximum compression lasts for only one frame before exploding upward and outward with vertical smears, shaking outlines, and exaggerated overshoot motion. Every element vibrates before settling into a completely new position.
[9–10.5s]
Textures migrate between forms. Dry rough textures sharpen previously fluid shapes into angular structures. Liquid textures soften rigid forms until they sag and drip. Luminous colorful textures dissolve into the background, causing the environment itself to pulse rhythmically. Every object changes behavior according to the texture it inherits.
[10.5–12s]
A massive rotational force sweeps across the frame. Every form flips, inverts, or turns inside out, exposing hidden reverse sides possessing qualities of hands, hooks, flames, or waves. Every collision triggers escalating chain reactions, transforming the entire visual world into an entirely new emotional reality.
[12–13.5s]
All boundaries dissolve. Forms merge, repel, and merge again, generating hybrid entities that inherit the colors and movement of their predecessors while becoming completely different beings. The background pushes forward into the foreground, becoming part of the forms themselves while releasing new shapes from within.
[13.5–15s]
Every previous process returns simultaneously at maximum intensity: contour separation, color bleeding, texture migration, and shadow-leading motion. Every frame is densely packed with intentional visual impact, fully absorbing and reinterpreting every reference element introduced earlier.
---
[15–22.5s: Fragmentation and Neurotic Storm]
[15–16.5s]
The storm collides with an invisible barrier. The center explodes into countless shards of water-covered broken glass. Within each fragment, glowing spiral irises rotate violently, generating centrifugal force that stretches and tears at the shard edges. Neon magenta and electric blue leak through every crack.
[16.5–18s]
The fragments fuse into overlapping collaged human faces. The faces are composed of purple stains, scratched paint, and rough pigment. Their expressions oscillate continuously between terror and ecstasy. Jagged lightning-like lines pierce through them like exposed nerve fibers, violently stitching the faces together.
[18–19.5s]
Color saturation reaches its peak. Electric blue lines solidify into physical cables stretching throughout space, twitching uncontrollably while releasing neon magenta sparks. The compressed paper background burns away, transforming into a charred, granular carbon-black surface.
[19.5–21s]
Gravity reverses. Liquid elements such as water droplets and wet ink surge upward while broken glass continues falling downward, producing intense visual tearing. Vertical smears appear along the edges of the frame as though an unseen force is physically wiping the entire animation off the canvas.
[21–22.5s]
Every form is pulled into countless miniature spiral-iris black holes. The irises rapidly expand until they dominate the entire screen, spinning at an increasingly suffocating speed. The rough tactile quality of chalk and dry brush textures contrasts violently against the brilliant electrical center, creating overwhelming sensory overload.
---
[22.5–30s: Aftershock and Eternal Echo]
[22.5–24s]
The giant iris evolves into pulsating biological tissue. The image breathes with a neurotic rhythm. Hidden symbols and textures appear and disappear beneath muscular contractions. Paint seepage creates convincing depth, as though layered across multiple transparent sheets of film.
[24–25.5s]
The animation enters an intense high-frequency vibration. Every object freezes in place while its edges oscillate violently like strings stretched to their breaking point. Violet shadows crawl across every surface, attempting to consume the remaining light.
[25.5–27s]
Rainbow-colored glass fragments and water droplets sweep across the frame like rain, refracting and distorting everything behind them through lens-like optical effects. Colors become heavy fluid masses that sink downward, leaving behind anxious, restless linear marks.
[27–28.5s]
The entire composition collapses into a single glowing horizontal jagged line. The background becomes a dark neon field filled with coarse grain. The line contains all accumulated energy, twisting violently while alternating flashes of electric blue and neon magenta pulse across it.
[28.5–30s]
The glowing line disintegrates into hand-drawn dust particles that drift rhythmically across the screen before gradually reassembling into one enormous blurred luminous silhouette. The final frame ends with tiny blocks of color continuing to vibrate subtly, leaving behind an everlasting emotional aftershock.
High-quality 3D animated cartoon, stylized modern family-animation film
look, oversized heads, huge expressive eyes, exaggerated proportions,
smooth rounded clay-like 3D surfaces, subsurface scattering skin, vibrant
saturated colors, warm cinematic animated lighting. NOT realistic, NOT
live action. 30 seconds, snappy comedic pacing.
LOCATION: a bright office break room, cheerful cartoon kitchen counter,
a small round table, an open fridge door behind them. On the table sits a
white plate holding nothing but crumbs and a smear of chocolate frosting.
CHARACTERS (locked, no drift):
MAYA — woman, oversized round head, curly dark hair, green blazer, huge
suspicious eyes, arms folded. Speaks English.
DIEGO — tall skinny man, light blue shirt, enormous eyebrows, very
dramatic hand gestures. Speaks Spanish.
LUC — short round man, grey knit sweater, sweet innocent face, a tiny
smear of chocolate frosting on his cheek that he never notices. Speaks
French.
COLOR 60:30:10 — 60% bright clean office whites and warm wood / 30% soft
mint-green wall / 10% accent of Maya's green blazer and the chocolate
frosting.
0–4s: wide shot. All three stand around the table staring down at the
empty plate in dramatic silence. Slow push-in on the crumbs like a crime
scene. Tense low strings.
4–8s: MAYA leans in, eyes narrowing, and says in English: "Okay. Which
one of you ate the last slice?" Cut to her suspicious glare.
8–13s: DIEGO throws both hands up theatrically, offended, and says in
Spanish: "¿Yo? ¡Yo ni siquiera como azúcar!" He clutches his chest,
scandalised.
13–18s: LUC blinks innocently, shrugs with total sincerity, and says in
French: "Moi non plus. Je suis complètement innocent." The frosting is
clearly visible on his cheek. Diego slowly turns and stares at him.
18–23s: MAYA points at Luc's face without a word. Beat of silence. LUC
touches his cheek, looks at his finger, and freezes, eyes going enormous.
23–27s: LUC, quietly, in French: "...Bon. J'ai peut-être goûté un petit
morceau." Maya and Diego both explode in overlapping outrage.
27–30s: Luc slowly licks the frosting off his finger anyway, completely
unrepentant, as the other two throw their arms up. Freeze on his guilty
little smile.
CAMERA: slow push-in on the plate, then natural shot-reverse-shot coverage
cutting to whoever speaks, a wide two-shot on the accusation, tight
close-up on Luc's frozen face, wide comedic final shot on all three.
AUDIO: tense low comedic strings during the standoff, a dramatic sting on
Maya's question, cartoon gasp from Diego, dead silence during the pointing
beat, overlapping outraged shouting at the end, one cheeky little
xylophone note on the final lick. Each character's lips match their own
line and language precisely. No background music over the dialogue.
FORBIDDEN: no auto-generated subtitles, no background music bed, no voice
crossing to the wrong character, no design drift, no realistic or
live-action rendering, no named studio references, no readable text or
logos.2f:T10a1,方向性 甲冑と馬のディテールを高速で見せ
type: anime_opening_continuous_camera_flythrough style: visual: dark fantasy anime, cel-shaded 2D, high contrast, full-color Japanese anime motion_quality: 24fps, 作画枚数多め, メリハリのあるスピードランプ tone: stylish, dramatic, kinetic, elegant, ominous direction: \u003e 日本のフルカラーアニメのオープニング。 高速なカット割りの代わりに、一筆書きのような連続カメラワークで、 複数のキャラクターを次々と捉えていく。 カメラは右へ左へ、上下へ、奥へ手前へと飛び回るローラーコースターPOVで、 一切止まらず、各キャラクターの最も映える瞬間を連続的につないでいく。 各キャラクターのポーズ・表情・背景・空間演出は、 そのキャラクターの性格、武器、属性、感情の核、物語上の役割から 最も魅力的に見える形をAIが自律的に判断して構成すること。 camera: motion: fast forward dolly with lateral drifts, rollercoaster POV, fluid aerial movement core_rule: \u003e no hard cuts, no static frames, no fades. 全編を通して一続きのワンカットとして成立させること。 speed_ramps: \u003e キャラクターのリビール直前に減速し、決めの瞬間で一瞬だけ視認性を確保し、 通過と同時に再加速する。 lens_feel: dynamic anime cinematography, dramatic perspective shifts, strong parallax final_exception: \u003e ラストのクライマックスキャラクター、または集合決めカットの直前のみ、 重みを出すために初めてカメラの運動を極端に緩める、 もしくは短く静止してもよい。 environment_logic: base_space: 無限に広がる異空間ヴォイド variation_rule: \u003e 各キャラクターのリビール区間ごとに、そのキャラクターの属性、感情、 象徴、戦闘スタイル、役割に最も合う環境要素をAIが選んで構成すること。 同じ空間を繰り返さず、毎回異なる空間的印象をつくること。 ただし全体としてはひとつの連続した異空間を飛行している感覚を保つこと。 palette_options: - deep crimson (#5A0000–#8B0000) + cyan mint accents + black shadows - midnight blue void + white petals + silver light - blood red mist + black feathers + gold rim light - violet smoke + shattered glass + cold cyan - 上記をベースに、各キャラクターに合わせて自由に再解釈してよい floating_elements_pool: - petals - feathers - blades - ribbons - chains - shattered glass - ash - sparks - torn paper - ink splashes - crystals - smoke - water droplets - leaves lighting: \u003e dramatic rim light, deep shadows。 光源設計はキャラクターの性格や役割に応じて変えること。 支配的なキャラには逆光や王者のような縁光、 儚いキャラには拡散光、 狂気のあるキャラには不安定な明滅光など、 感情の印象を光で補強すること。 input_characters: description: \u003e 以下にキャラクターのリストを与える。 人数は固定しない。2人でも10人以上でも対応可能。 各キャラクターについて、名前・性格・武器・属性・役割・感情の核・象徴モチーフなどを参照し、 そのキャラらしさが最大化する一瞬を構成すること。 format_example: - name: Character Name role: protagonist / rival / villain / mentor / key figure / support personality: "" weapon_or_power: "" attribute_or_theme: "" emotional_core: "" symbolism: "" priority: high / medium / low reveal_planning: global_rule: \u003e 入力された全キャラクターを、物語の高まりが自然に増していく順に並べ替えてよい。 基本的には、軽め・不穏・謎めいた導入から始まり、 中盤で躍動感と情報量を増し、 終盤で主人公・対立軸・ラスボス・最重要人物へ向けて熱量を上げること。 per_character_rule: \u003e 各キャラクターごとに以下を自動で設計すること: 1. pose: 静止ではなく「動きの最中の一瞬」 2. expression: 感情の核が伝わる表情 3. framing: 最も映えるカメラ距離・角度 4. background_variant: そのキャラに対応した空間演出 5. foreground_wipe: 次のキャラへつなぐ前景通過物 6. motion_relation: カメラに対して向かってくる / すれ違う / 見下ろす / 落下する / 回転する 等 uniqueness_rule: \u003e 各キャラクターで、体勢、シルエット、画面構図、視線方向、背景モチーフ、 前景の抜け、動きのベクトルを被らせないこと。 連続して似た演出を使わないこと。 duration_rule: \u003e 各キャラクターの見せ尺は均一でなくてよい。 priority が高いキャラクターほど、わずかに長く、印象的に見せること。 低優先キャラは短く鋭く見せ、高優先キャラは減速を強めて存在感を出すこと。 character_processing_loop: instruction: \u003e 入力されたキャラクター配列を上から順に機械的に処理するのではなく、 オープニング映像として最も気持ちよく流れる順番に再構成してよい。 それぞれについて以下を判断して映像化すること。 for_each_character: decide: - このキャラは支配的か受動的か - 攻撃 / 防御 / 逃走 / 佇み / 観察 / 詠唱 / 指揮 のどれが最も似合うか - 武器や能力をどう一瞬で象徴的に見せるか - 視線、手、指、髪、衣装、武器のどこを強調すべきか - カメラがそのキャラに向かうべきか、横切るべきか、追い抜くべきか - そのキャラに最も合う背景と前景モチーフは何か output: - pose - expression - framing - background_variant - foreground_wipe - transition_vector transitions: between_reveals: \u003e 花弁、羽根、刃、リボン、鎖、ガラス片、煙、灰、インクなどが 必ずレンズ手前を至近距離で横切り、それをワイプ代わりにして次のキャラへつなぐこと。 rule: \u003e 必ず「前景がレンズを横切る → 空間が変化する → 次のキャラが現れる」 という流れを守ること。 continuity: \u003e 前景ワイプは単なる切り替えではなく、 カメラが本当に次の空間へ飛び込んでいく感覚を生むように使うこと。 climax_logic: final_focus_rule: \u003e 最後の区間では、最重要キャラクター (主人公 / ラスボス / キーパーソン / 物語の中心人物)に到達すること。 そのキャラは他よりも重く、強く、象徴的に見せること。 ending_options: - 最重要キャラクター単体でカメラが初めて止まり、一枚絵として決まる - 主人公の先に最大の敵が現れ、緊張感のある静止に入る - 最重要キャラのあと、短く複数人の集合構図へ流れ込み、一枚絵で締める final_image_rule: \u003e ラストはポスターのように成立する構図にすること。 それまでの流動性と対比して、最後にだけ「止まる意味」があるように演出すること。 output_goal: \u003e 結果として、全体が一本の連続したアニメOPショットとして成立し、 各キャラクターが短時間でも明確に識別でき、 しかも一人ずつ違う魅力で刻み込まれること。 単なる紹介映像ではなく、「この作品世界の熱量と因縁」が伝わること。 negative: - hard cuts - static camera (except final climax hold) - fades - text - logos - subtitles - photorealism - 3DCG look - repetitive poses - repeated environments - flat staging4e:Ta6b,Full prompt:
POV — Morning Commute in Coastal Japan | Ultra-Realistic, One Continuous Take
First-person POV, no cuts, one unbroken handheld-feeling shot from our own eyes on the way through a small seaside town. Ultra-photorealistic, phone-in-hand realism — natural camera micro-shake, subtle head bob with each step, occasional glance down at our own hands and shoes. 4K, shallow but natural depth of field, real morning light, no color-grading fantasy, no neon.
Beat 1 — Descending into the station (0–4s): We walk down worn concrete station stairs into a small coastal-town metro entrance, one hand loosely gripping a canvas bag strap that swings into the lower frame. Fluorescent tint mixes with daylight bleeding from the entrance behind us. Commuters pass in soft focus; we hear muffled announcements and the shuffle of footsteps.
Beat 2 — The orange cat (4–9s): We slow and crouch — the horizon dips as our POV lowers toward the ground. An orange tabby sits by a ticket-gate pillar, tail curled. Our hand (natural skin texture, slightly bitten nails, cuff of a sleeve) reaches into frame and gently strokes its head. The cat leans into the touch, eyes half-closing, ears flicking. We hold there a beat, warm and quiet.
Beat 3 — Boarding, settling by the window (9–14s): We rise, the horizon lifts, and we step through the train doors just as the chime sounds. POV turns and moves toward a window seat; we sit, and the frame settles against the glass. A faint silhouette of us ghosts over the view. The train pulls away with a gentle lurch.
Beat 4 — The shimmering sea (14–20s): Through the window, the town gives way to open coastline. Sunlight scatters across the water in thousands of moving sparkles — real specular glitter, not CGI bloom. Utility poles and wires strobe past in the foreground. We rest our chin lightly (subtle downward frame tilt), just watching.
Beat 5 — Rails into the sea (20–25s): The track curves and runs so close to the shore that the rails seem to vanish straight into the water — the sea fills nearly the whole frame, waves lapping right up to the ballast. Foam catches the light. We're gliding along the edge of the ocean, the horizon line steady, gulls crossing in the distance. Hold on this until the shot ends.
Technical spec: continuous POV, first-person, no cuts or transitions, natural handheld micro-motion, realistic morning ambient audio (station hum, cat, train chime, rail clatter, distant surf), authentic Japanese coastal-railway setting (Shimonada / Enoden-style seaside line), overcast-to-clear soft daylight, photoreal skin and fabric texture, 2.39:1 or 16:9, ultra-realistic.4f:T12c8,@dreamina_ai T2V prompt: A 30 second ultra realistic handheld POV video set in France at around 3 am, outside a busy bar and nightlife street. The camera is the point of view of a young woman filming one handed on an iPhone. We never see the phone itself at any point. The footage must feel like raw, authentic amateur nightlife footage with natural iPhone aesthetics.
It is very late. People are leaving bars, smoking outside, waiting for Ubers, talking too loudly, eating kebabs, laughing in tired, sloppy groups. The mood in the street is a mix of drunken energy and visible exhaustion. Faces look tired, makeup is slightly worn, outfits are a bit messy after a long night. The pavement is slightly wet and reflects neon signs, headlights, and warm bar lights. The air feels cold and heavy. The whole scene must feel intensely real, socially awkward, and uncomfortable.
The camera operator is a woman, and her friend is another young woman walking a few steps ahead of her, visibly angry. Th
Shot 1 (2.0s): An extreme close-up on @Dragon's closed eye as it suddenly snaps open, the golden iris contracting sharply. The camera holds tight and still, dust particles drifting past in the beam of light. Sharp, sudden shift in tension with a subtle lighting flicker.
Shot 2 (2.5s): A quick, reactive medium shot as @Boy stumbles backward in fear, nearly tripping over rubble, his expression panicked. The camera shakes handheld, following his frantic retreat. Lighting shifts to a colder, harsher tone as shadows stretch across the hall.
Shot 3 (3.0s): A powerful low-angle wide shot as @Dragon rises up, wings spreading and casting a massive shadow over the hall. It lets out a thunderous roar. The camera tilts up rapidly to capture its full towering height, dust and debris shaking loose from the ceiling. Dramatic high-contrast lighting with the beam from above now cutting harshly across the scene.
Shot 4 (2.5s): A fast-paced action shot as @Dragon swipes its massive claws toward @Boy, who dives and rolls out of the way across the stone floor. The camera whip-pans to follow the motion, adding kinetic energy. Debris scatters, lighting flickers with dust kicked into the air.
Shot 10 (2.5s): A dynamic tracking shot as @Boy scrambles behind a fallen pillar for cover, breathing heavily. @Dragon rears back, preparing to attack. The camera circles quickly around the pillar, low to the ground, building tension. Cold blue-white lighting dominates the frame.
Shot 11 (3.0s): A dramatic wide shot as @Dragon unleashes a burst of icy breath across the hall, frost spreading rapidly over the stone floor and pillars. @Boy sprints and slides across the ground to dodge the blast, ice crystals forming in the air around him. The camera tracks low and fast alongside his movement, capturing the chaos.
Shot 12 (2.0s): A final tense close-up as @Boy presses his back against a frozen pillar, chest heaving, staring up at @Dragon towering above him. The camera holds still, slowly pushing in. Cold light reflects off the newly formed ice, creating a striking, high-contrast final frame.
High-quality 3D animated cartoon, stylized modern family-animation film
look, oversized heads, big expressive eyes, exaggerated proportions,
smooth rounded clay-like 3D surfaces, subsurface scattering skin, vibrant
saturated colors, cinematic animated lighting. NOT realistic, NOT live
action.
Single continuous 30-second shot, no cuts, one flowing camera move.
SUBJECT: a young woman with an oversized round head, huge expressive
eyes, dark hair in a messy bun, a long mustard-yellow raincoat, boots, a
small satchel. Same design, hair, and outfit throughout, no drift.
LOCATION: a cozy stylized night market in the rain, narrow covered lane,
strings of warm hanging bulbs, steaming food stalls with rounded cartoon
awnings, friendly cartoon vendors and shoppers moving around her, glossy
wet cobblestones reflecting the lights, puddles everywhere.
COLOR 60:30:10 — 60% deep blue rainy night and wet stone / 30% warm
golden bulb and stall light / 10% accent of her mustard-yellow raincoat.
0–8s: the camera tracks backward ahead of her as she walks through the
market with a small bounce in her step, dodging puddles, steam curling
from stalls, vendors waving as she passes.
8–16s: the camera arcs smoothly around to her side, revealing the full
depth of the market behind her, then rises to follow just over her
shoulder as she stops to smile at a stall of glowing paper lanterns.
16–24s: she turns down a quieter narrow alley, the camera following, the
crowd noise falling away, single bulbs swinging gently overhead, her
reflection rippling through a puddle as she steps over it.
24–30s: the camera cranes up and pulls back into a wide as she reaches a
warmly lit doorway at the alley's end, pauses, shakes the rain off her
coat, and steps inside. Hold on the glowing doorway.
PHYSICS: raindrops bounce and splash with cartoon weight, her coat swings
and ripples as she walks, puddles ripple outward from each step, steam
rises and curls, hanging bulbs sway gently.
AUDIO: rain pattering on awnings, cheerful market chatter, sizzling food,
her boots splashing, the noise thinning to soft rain in the alley, a warm
door chime at the end.
FORBIDDEN: no cuts, no design drift in face, hair, or coat, no realistic
or live-action rendering, no auto-generated subtitles, no background
music, no readable text or logos, no named studio references, no warping
geometry.75:T783,Ma
30-second cinematic anime action sequence, breathtaking high-budget anime film, ultra-dynamic editing, fluid sakuga animation, spectacular fight choreography, seamless kinetic camera movement, aggressive speed ramps, rapid match cuts, whip pans, rotating camera moves, dramatic perspective shifts, cinematic lens distortion, volumetric lighting, dust, sparks, debris, explosive impacts, emotionally intense, premium animation quality, no dialogue.
0–5s
A ruined futuristic city stands silent beneath storm clouds.
Two lone warriors face each other hundreds of meters apart.
The camera rapidly alternates between extreme wide shots, close-ups of determined eyes, hands tightening around weapons, feet scraping broken concrete, fluttering coats and drifting dust.
A heartbeat.
Absolute silence.
5–10s
Both fighters explode forward simultaneously.
The camera races alongside them before instantly switching to overhead drone views, impossible low angles beneath their feet, first-person perspectives, and ultra-fast lateral tracking shots.
Shockwaves ripple across the ground as they collide.
The first impact sends debris flying in every direction.
10–15s
An overwhelming combat montage.
Lightning-fast sword strikes.
Acrobatic flips.
Mid-air clashes.
Wall running.
Buildings collapse around them.
The camera never stays still: continuous orbit shots, rapid whip pans, rotating 360-degree movements, crash zooms, Dutch angles, seamless transitions through clouds of dust and sparks.
Constant speed ramps amplify every decisive strike.
15–20s
The battle escalates.
Energy erupts with every collision.
One fighter launches the other through multiple rooftops.
The camera follows in a single uninterrupted shot before cutting to an extreme close-up as both instantly reappear, colliding again in mid-air.
Fragments of glass, steel and concrete freeze momentarily before exploding outward.
20–25s
The choreography becomes almost impossible to follow.
Dozens of rapid exchanges.
The camera jumps between ultra-wide cinematic compositions, intimate facial close-ups, spinning aerial shots, slow-motion impacts, handheld ground-level tracking, and impossible anime-style perspective shifts.
Every cut increases the intensity.
The music reaches its peak.
25–30s
Everything suddenly slows.
The two warriors cross paths one final time.
Silence.
A single decisive strike.
The camera follows the blade in extreme slow motion before rapidly pulling back into an enormous aerial shot revealing the devastated city beneath dark skies.
Dust slowly settles.
Hard cut to black.
Sound design: colossal cinematic orchestra blended with modern hybrid percussion, taiko drums, distorted synth pulses, thunderous impacts, rushing wind, metallic blade clashes, explosive shockwaves, collapsing buildings, debris showers, aggressive cinematic whooshes, deep bass hits, brief silence before the final strike, overwhelming orchestral finale with massive surround sound.7c:T1509,Use
1Write it in one fixed order. Subject, then action, then location, then camera work, then lighting, then picture texture. "A person walking" gets you a stock loop; the same subject with a camera move and a light direction attached gets you a shot. Creators who have spent real credits on 2.5 converge on this order.
2Thirty seconds is written as stages, not as longer sentences. The official guide splits a long clip into [STAGE ONE] / [STAGE TWO] / [STAGE THREE], each with an opening state, one main event, and an end state the next stage inherits. Padding a 10-second prompt with adjectives does not make a 30-second clip; splitting the event into inherited states does.
3One main action per time segment, described exactly once. Whether you use stages or plain 0-5s / 5-10s ranges, give each segment a single main movement. Two things go wrong otherwise: three actions crammed into five seconds all get compressed, and the guide is explicit that describing the same action twice degrades it.
4Give every reference asset a job and an exclusion. Not "here are my images" but "@image1 supplies the armour design and colour breakdown. Do not take the background." You can attach up to 50 assets; the more you attach without assigning roles, the faster stability drops.
5Name your subjects and reuse the name. Write <SOLDIER>, <SHARI>, <WOMAN> once and refer to that token every time after. Over 30 seconds, "she" and "the man" drift onto the wrong body — a named token is the cheapest identity lock there is.
6Write the sound as a cue sheet, not as a mood. Audio comes out of the same pass as the picture, so list what you want: ambience, the specific foley, dialogue with the language named, and what should be absent. Several prompts in this library end with a full sound list, and it is a large part of why they look finished.
7Ask for the imperfections you want. Handheld shake, autofocus that settles late, exposure breathing, framing that clips a face. Realism in this model is a list of specified flaws — leave them out and you get a clean commercial no matter how many times you write "candid".
8End with the negative list, and keep parameters out of it. Close with what must not appear: no subtitles, no watermark, no extra limbs, and — for reference-driven jobs — never render the reference sheet itself or duplicate the subject. Resolution, duration and aspect ratio are generation settings, not prompt text.
Prompt formula
Subject + action + location + camera work + lighting + picture texture + sound. Past about ten seconds, wrap that in timed stages: each stage gets one main action and an end state the next stage inherits, then close with the negative list.
Example: <BARISTA> pulls the last shot of the night in a closing café and sets the cup on the pass. Rain on the window, chairs already stacked, one warm pendant lamp over the machine and the rest of the room in shadow. The camera holds a low medium on the portafilter, then rises to a profile as the cup lands. Photoreal, warm tungsten against cold window light, fine grain. Sound: the grinder winding down, steam hiss, rain, one mug on wood, no music. Do not add customers, subtitles or a watermark.
Frequently asked questions
What is Seedance 2.5?
Seedance 2.5 is ByteDance's video generation model, unveiled on 23 June 2026 at the Volcano Engine FORCE conference. On the shipped API it generates 4 to 30 seconds in a single pass (or duration -1 to let the model pick the length), takes up to 50 multimodal reference assets in one job (up to 30 images, up to 10 video clips and 10 audio clips), adds video editing and video extension as task types so a product or background can be swapped without re-rolling the whole clip, and generates audio in the same latent space as the picture. Output is 480p or 720p — the native 4K at 10-bit colour announced at FORCE is not something the API exposes. ByteDance also shipped an official prompting guide for it, which we have walked through in English.
Are these prompts free to use?
Yes. Every prompt on this page is free to copy and run anywhere you have access to Seedance 2.5 — no login, no payment.
Who wrote these prompts?
Two kinds, and they are labelled differently. Prompts the creator published are kept in the language they wrote them in, credited, and linked to the original post — copyright stays with the author. For clips whose creator never published a prompt, we reverse-engineer one from the finished video and mark it "reconstructed": it describes the output, so it cannot recover negative constraints, exact dialogue or the reference-image workflow. It is a writing reference, not the creator's prompt. If you are an author and want an entry removed or a credit corrected, email us.
Where can I run Seedance 2.5 today?
On apimodels.app, as the model seedance-2.5, since 8 August 2026 — 480p $0.134/s, 720p $0.300/s, charged only on success. That is a change: probing our production Volcengine Ark account on 4 August 2026, every Seedance 2.5 model id still returned InvalidEndpointOrModel.NotFound, and we said so rather than sell a page that errors on generate. One key covers it alongside Seedance 2.0 (still the one to use if you need 1080p, from $0.092/s), MiniMax H3, Kling, VEO and Grok. The creators in this library ran these prompts in ByteDance's own Dreamina (Jimeng) and in Higgsfield.
How is writing for Seedance 2.5 different from Seedance 2.0?
The craft is the same; the structure is not. At 4-15 seconds you can write one paragraph for one action. At 30 seconds you have to split the clip into stages, give each stage one main action and an end state the next stage inherits, and lock identity with named subject tokens instead of pronouns — otherwise the model spreads the action evenly and identity drifts across the back half. The reference workflow also scales: 2.0 takes up to 9 images plus a reference video and audio, 2.5 takes up to 50 assets, which makes assigning each one an explicit job and exclusion much more important.
How long should a Seedance 2.5 prompt be?
As long as the clip needs and no longer. The prompts in this library run from about 1,700 to 4,600 characters, and the longest ones are the 30-second pieces — not because they are wordier, but because a 30-second clip has more states to pin down. If your prompt is long because it describes the same action three ways, cut it; the official guide is explicit that repeating an action degrades it.
Run these prompts on apimodels.app
Seedance 2.5 is live here alongside 85+ other image, video, audio and language models — one API, one key, pay as you go, $1 free on sign-up.
The prompts we reconstructed are published in full under MIT; author-written prompts are indexed with credit and a link to the original post. Open an issue to suggest more.