Full prompt
Model Comparisons
Model comparison: this source video includes Seedance 2.5 alongside other model outputs. The stored clip is not represented as exclusively generated by Seedance 2.5.
Full prompt
Model Comparisons
Model comparison: this source video includes Seedance 2.5 alongside other model outputs. The stored clip is not represented as exclusively generated by Seedance 2.5.
Create with Seedance 2.5
Adjust the subject, action, camera, or timing, then start a new Seedance 2.5 generation.
Seedance 2.5
Your generated video appears here
Write or load a prompt, review the live credit estimate, and start the generation when the settings look right.
More Seedance 2.5
Model Comparisons
<<<Image1>>> は Girl A の顔、黒髪の肩上ボブ、青い瞳、体格、制服デザインを定義する。Girl A は標準語を話す真面目な吹奏楽部員で、担当楽器はクラリネット。白背景や元のポーズは継承しない。 <<<Image2>>> は Girl B の顔、茶髪ポニーテール、体格、制服デザインを定義する。Girl B は関西弁を話す明るい吹奏楽部員で、担当楽器はトランペット。白背景や元のポーズは継承しない。 <<<Image3>>> は Girl C の顔、ダークブラウンのボブヘア、体格、カーディガンを含む制服デザインを定義する。Girl C は博多弁を話すおっとりした吹奏楽部員で、担当楽器はフルート。白背景や元のポーズは継承しない。 この3枚の参照画像は3人の別々の女子生徒を定義する。主要人物はこの3人のみ。顔、髪型、衣装、担当楽器、方言、性格を混同しない。 夏休みの高校の吹奏楽部を描く、爽やかで自然な日常アニメ映像。舞台は夏の朝の音楽室。窓の外は強い日差し、青空、校庭の緑。室内には譜面台、椅子、楽器ケース、扇風機、開いた窓、少し揺れるカーテン。外から蝉の声が聞こえる。空気は少し暑く、夏休みの学校らしい静けさがある。字幕なし、ナレーションなし。 全体のカメラ方針: 自然なマルチカット構成。1カットごとに目的を持たせる。引きの状況説明、寄りの仕草、手元の演奏、表情のリアクションを交互に見せる。人物を棒立ちにしないため、会話中も視線移動、重心移動、髪の揺れ、汗をぬぐう、譜面を見る、楽器を持ち直す、息を吸う、指を動かすなどの小さな動作を常に入れる。カメラは不自然に激しく動かさず、ゆるいドリーイン、軽いパン、短い横移動、自然なカットつなぎで構成する。演奏中は表情、指、口元、楽器、視線のやり取りを細かく切り替える。 全体の音: 蝉の声、教室の環境音、椅子を引く音、楽器ケースを置く音、扇風機の回転音、カーテンが揺れる気配、楽器を持つ音、短いチューニング音、息づかい、そして短い合奏。 0〜5秒 — 登場と状況説明 カット1: 音楽室の中を外の窓側から見たやや広めの引き。カメラはゆっくり前進する。朝の光が斜めに差し込み、窓辺のカーテンが揺れている。Girl A は譜面台の前で譜面を見ている。Girl B はドア側から入ってきて、少し前かがみで暑そうにしている。Girl C は楽器ケースを机に置く。 カット2: Girl A の胸上のミディアムショット。譜面から目を上げ、2人を見る。クラリネットの部品を持ちながら落ち着いて言う。 Girl A が標準語で言う: {おはよう。最初のフレーズ、合わせよう。} 終了状態:3人ともそれぞれの準備に入る空気になる。 5〜10秒 — 暑さと夏休み感 カット3: Girl B のミディアムショット。カメラは少し右から寄る。Girl B は制服の襟元を軽くつまみ、片手であおぐようにしながら、汗をぬぐって言う。ポニーテールが少し揺れる。 Girl B が関西弁で言う: {音楽室あっつ!} カット4: 扇風機の回転を一瞬映すインサート。続いてカーテンが揺れるカット。夏の空気を見せる。 カット5: Girl C のやや寄りのミディアムショット。Girl C は楽器ケースを開けながらふっと笑い、フルートを取り出して言う。 Girl C が博多弁で言う: {でも夏休みの部活って感じやね。} 終了状態:3人の気分がそろい、練習モードに入る。 10〜15秒 — 準備と合図 カット6: 手元の短いモンタージュ。テンポよく3つの短いカットをつなぐ。 - Girl A がクラリネットを組み立てる指先の寄り - Girl B がトランペットのバルブを軽く押して感触を確かめる寄り - Girl C がフルートを水平に持ち、キーに指を置く寄り カット7: 3人の半身が入る横並びのミディアム3ショット。カメラはゆっくり左から右へ小さくスライドする。3人が譜面台に向き直る。Girl A は譜面を見てから2人を見る。Girl B は姿勢を正し、Girl C は小さくうなずく。 カット8: Girl A の顔寄り。目線で合図し、小さく息を吸って言う。 Girl A が標準語で言う: {いくよ。} 終了状態:3人が同時に吹き始める直前。 15〜24秒 — 合奏メイン 演奏パートは会話よりも長く取り、部活らしい集中を見せる。カメラはテンポよく切り替えるが、空間の向きは崩さない。 カット9: 3人を正面から捉えたミディアム3ショット。合図と同時に演奏開始。最初の1〜2秒はこのカットで、3人が同時に入る瞬間を見せる。息を吸ってから音を出す流れを自然に見せる。 カット10: Girl A のクラリネットの手元と口元の寄り。指が動き、真剣な視線が譜面へ向く。 カット11: Girl B のトランペットのややローアングル寄り。ベルの輝き、頬の張り、バルブを押す指の動き、リズムに少し乗る体の揺れを見せる。 カット12: Girl C のフルートの横顔寄り。フルートの水平ライン、指先、落ち着いた息遣い、やわらかな目線を見せる。 カット13: 再び3人の斜め前からの3ショット。カメラがわずかにドリーインし、3人の音がまとまっていく感じを見せる。演奏中、Girl A は譜面と2人を交互に見て、Girl B は少し楽しそうに吹き、Girl C は安定して支える。 カット14: 短いリアクション挿入。Girl A の目線、Girl B の口元、Girl C の指先をリズムよく短く切り替えて、演奏の躍動感を出す。 終了状態:フレーズを吹き終え、音が止まり、3人が手応えを感じる。 24〜27秒 — リアクション カット15: 演奏直後の3ショット。音が止まり、一瞬だけ静けさ。Girl B が顔を上げて、嬉しそうに笑う。Girl A は少しほっとし、Girl C はうなずく。 カット16: Girl B の寄り。トランペットを少し下ろして、明るく言う。 Girl B が関西弁で言う: {今のええやん
Model Comparisons
Create a vertical 9:16 summer-night outdoor idol concert in a polished Japanese animated feature-film style. REFERENCE ASSIGNMENTS: Use@Image 1 as the definitive character and costume sheet for the solo female idol Maya Kisaragi. Faithfully preserve her long wavy pastel-pink hair, pink eyes, rose hair ornament on the left side, white-pink-and-gold rose-themed summer concert costume, jewelry and shoes. Use@Image 2 as the reference for Maya’s face, expressions, anime rendering quality and on-stage singing presence. Use@Image 3 as the definitive reference for the large rose-shaped outdoor stage, central floral LED screen, runway, pink-and-blue concert lighting and audience layout. Use@Image 4 as the design reference for the white and rose-gold rose-decorated microphone and stage props. Maya is the only main idol in the video. Do not interpret the front, side and back views in the character sheet as separate people. Preserve exactly the same face, hairstyle, body proportions, costume, accessories and microphone throughout the video. FIRST 15 SECONDS — MANDATORY SHARED SEQUENCE: 0–3 seconds: Begin at night from a low over-the-shoulder audience viewpoint. A full stadium crowd waves pink glow sticks. Smoothly dolly forward along the runway toward Maya standing at its center. Stage lighting shifts from pink to blue-violet in rhythm with the music. Rose petals drift naturally through the summer night air. 3–8 seconds: Continue the same forward camera movement into a medium shot. Maya sings brightly into the rose-decorated microphone in her right hand, with natural lip synchronization. She moves her left hand from her chest outward toward the audience and smiles warmly. Her long hair, ribbons and translucent skirt layers move naturally with her body and the light night breeze. 8–12 seconds: Without cutting, arc gently around Maya from left to right and move into a waist-up medium close-up. The giant rose LED behind her blooms with light while thousands of pink glow sticks form a deep wave of light in the background. Keep her face, hands, costume details and microphone stable and clearly rendered. 12–15 seconds: The chorus reaches its first climax. Pull back slightly to reveal Maya’s full body and the stage. While continuing to sing, she raises her left hand toward the night sky, then reaches toward the audience. One synchronized burst of pink-and-gold fireworks appears on both sides of the stage and the audience cheers. At exactly 15 seconds, form a visually complete climax that works as a natural ending for a 15-second video. CONDITIONAL CONTINUATION FOR VIDEOS LONGER THAN 15 SECONDS: If the selected video duration is 15 seconds or less, do not compress, preview, or mix any of the following events into the first 15 seconds. 15–21 seconds: Continue the same uninterrupted take. Maya walks gracefully down the runway while singing. The camera travels backward in front of her, keeping a stable medium-full composition. She looks toward the audience on both sides and gives a small wave. Synchronize her walking, singing, hair and costume motion naturally. 21–26 seconds: At the circular end of the runway, slowly orbit around Maya. Reveal the audience’s pink glow sticks as an ocean of light, with the rose stage and distant fireworks filling the vertical background. Maya smiles toward the camera, then points the microphone toward the crowd for a brief call-and-response. 26–30 seconds: The audience sings and cheers, and Maya performs the final phrase. Smoothly push into a frontal close-up. She makes eye contact with the camera, smiles gently and places her free hand over her chest. The rose LED glows brightly behind her as soft pink petals pass through the foreground. End on a clean, beautiful hero shot with her face clearly visible. VISUAL DIRECTION: Elegant, vibrant Japanese animated feature-film quality; delicate linework, luminous colors, soft skin shading and a pink, rose-gold and blue-violet palette. Preserve facial and costume details under strong concert backlighting. Use smooth cinematic camera motion, natural spatial depth and energetic live-concert atmosphere. Keep Maya inside the central safe area of the vertical frame. Do not crop her face, expressive hand or microphone at important moments. AUDIO: Bright and emotional Japanese idol pop with female lead vocals, live drums, bass, guitar and sparkling synthesizers. A youthful, clear Japanese singing voice. Synchronize the vocals, lip movement, band, choreography and lighting changes. Include natural crowd cheers and calls, spacious outdoor concert ambience and the low distant impact of fireworks. Keep the song dominant and the crowd below the vocal level. No clipping, distortion or random noise.
Model Comparisons
At 0.0s, [0s-4s] the wide tracking shot follows the teenage girl as she rapidly pedals her vintage dark green step-through bicycle downhill along the weathered asphalt road, with the camera facing the east vector while violent coastal winds whip her hair backward against a bright blue sky. She is wearing a wide-brimmed straw hat with a red ribbon, a white sailor summer shirt with a blue collar, a navy blue pleated skirt, and white canvas shoes, maintaining strict structural permanence of her wardrobe. [4s-8s] The camera smash cuts to a low-angle extreme close-up macro tracking shot as the rapidly spinning front tire violently slices through a shallow rain puddle on the rough asphalt, blasting a dynamic spray of water droplets directly toward the camera lens. [8s-11s] The camera violently whip-pans up to a medium close-up of her side-profile as a sudden heavy gust of wind from the west vector ocean horizon blows her straw hat backward, forcing her to squint her eyes and smile peacefully. [11s-15s] The camera smoothly cranes up from behind her back in an over-the-shoulder wide reveal shot as she firmly brakes to a halt in the foreground, revealing a vibrant yellow-and-green commuter train roaring horizontally across a rusted railway crossing in the deep background. The entire scene takes place in a strict 100% Euclidean real-world outdoor environment featuring a clean separated asphalt pavement surface and a raised concrete curb physically separating the road from overwhelmingly lush tall green summer grass, ensuring zero melted roads and zero phantom trains. The visual execution is a cel-shaded Studio Ghibli anime aesthetic featuring intense triadic color separation of the bright blue ocean, vivid green grass, and the yellow train, enhanced by interactive warm sunlight wrapping around the subject's cheekbones and thick volumetric atmospheric summer haze, completely free of plastic 3D CGI slop or character morphing.
Model Comparisons
Using Higgsfield — epic cinematic fantasy-sci-fi battle, one continuous 30-second shot with smooth multi-angle flow and tempo shifts, no hard cuts, no UI. A lone glowing sorceress duels a colossal armored titan-mech across a world of floating jungle cliffs. Heroine: A slender warrior-sorceress wreathed in radiant emerald-green energy — flowing luminescent robes, glowing vein-like markings across her skin, hair lifting in the power. Defiant, hands raised, channeling green light. Fully original character design, consistent across every shot. The Titan: A colossal crimson-and-obsidian armored mech-titan, sleek beveled plating with glowing seams, towering over the canyon. It winds up and unleashes a massive blinding energy beam. Original mecha design, not based on any franchise. World: A breathtaking alien landscape of floating jungle-topped cliffs and rock spires wrapped in mist, cascading waterfalls, lush canopy, drifting debris and embers, giant pale planets low on the horizon, volumetric god rays through haze. Enormous, awe-inspiring scale. Shot flow (continuous, varied angles): 0–5s — Sweeping aerial glide over the floating cliffs and waterfalls as a first lance of energy tears across the canyon, embers streaming. 5–11s — Camera drops to the sorceress on the cliff edge, emerald power flaring, a tight low hero angle as the titan looms through the mist before her. 11–18s — Close intensity: her glowing markings pulse, eyes locked; cut of motion to the titan winding back, servos groaning. 18–25s — The titan fires a massive golden-red beam across the chasm; she braces behind a shield of green light, rock shattering, dust exploding, camera shaking with the impact. 25–30s — The beam fades, dust settles on the reeling giant, the sorceress still standing; camera pulls back for an epic hero-wide of the whole vista. Motion & FX: enormous scale and weight, cinematic slow-fast pacing, volumetric dust and ember sims, energy-beam bloom and light scatter, debris physics, atmospheric haze, dynamic sweeping camera, strong shot-to-shot character consistency. Style: hyper-realistic cinematic VFX, epic blockbuster fantasy-sci-fi, Unreal Engine 5 / film-grade render, ray tracing, HDR, rich color grade (emerald green vs crimson-gold), anamorphic lens flares, shallow depth of field, film grain, 4K, 24fps. Audio (native): deep cinematic score swell, titan servo groans, crackling energy, the roar and boom of the beam, wind. NEGATIVE PROMPT: UI, HUD, game overlay, on-screen text, watermark, logos, celebrity likeness, movie or franchise IP, copyrighted characters, cartoon, low resolution, blurry, distorted anatomy, extra limbs, fused fingers, warped mech, morphing geometry, inconsistent character between shots, flickering, washed-out colors, jitter
Model Comparisons
A 20-second 21:9 ultra-widescreen plush universe adventure, rendered in hyper-detailed plush materials, cinematic cosmic scale, cute forms contrasted with intense action, exaggerated FPV camera work, and authentic soft-body physics. A spaceship made of plush fabric, buttons, zippers, and stuffing carries three small plush animals—a rabbit, a fox, and a crow—fleeing at high speed through a universe of giant yarn planets, fleece nebulae, and stuffed-toy beasts. A giant interstellar whale covered in deep-blue long plush fur with glass-button eyes chases them from the nebula. The action includes the zipper-ship catapulting into flight, low-altitude high-speed orbit around a yarn planet, the whale bursting out behind, weaving through a button asteroid belt, being swallowed into the whale's body, escaping through a golden zipper on its back, the whale slowly deflating like a toy losing its stuffing, and finally bursting through a pink-purple fleece nebula toward a colorful button sun.
Model Comparisons
Create a premium 2D motion-graphics brand film for the AI brand "Here We Are" — strictly 15 seconds, 16:9 landscape, 4K, 60fps. The overall style combines contemporary art publications, Riso stencil printing, hand-torn paper collage, magazine cut-outs, Swiss editorial design, and high-speed experimental kinetic typography. The frame is dominated by an off-white paper background, using deep ink black, signal blue, coral orange, and soft lilac. The core concept is "From here, we become": a large number of small dots of varying kinds are initially scattered across the frame, then rapidly sense, probe, chase, approach, collide, connect, cluster, swap positions, and synchronize — finally forming "We" out of multiple instances of "Here," then establishing a continuously holding "Are," and organically generating the brand name "Here We Are." Three through-lines must hold for the entire film. First, the dot swarm keeps moving from open to close and must never stay still for long. Second, paper tearing, cutting, page-turning, overprinting, and layout recomposition are always involved in transitions and must never degrade into clean generic vector motion graphics. Third, every dot, paper strip, arc, and motion path must correspond to the real letterform stroke structure of Here, We, and Are — it must never become a random dot network, meaningless capsules, or generic geometric animation. On average, a clearly visible change must occur every 0.35 to 0.65 seconds, with at least one strong rhythmic action per second, including high-speed stamping, sudden acceleration, elastic collision, paper pull-away, massive scale jumps, direction reversal, brief letterform resolution, hard layout cuts, or torn-paper occlusion. Slow drifting, constant-velocity sliding, prolonged static typography, and soft cross-fades are forbidden. "Here," "We," and "Are" may appear briefly and accurately in the middle section, and "Here We Are" must appear accurately at the end. Apart from these, no other readable text, numbers, URLs, labels, garbled characters, watermarks, or platform logos may appear. All dots must be completely flat 2D solid paper pieces — no spheres, speculars, shadows, reflections, glass, metal, volumetric gradients, or perspective thickness. The paper pieces have a clear handmade collage texture: slight torn fibers, scissor cuts, rough paper edges, uneven ink, halftone dots, tiny print specks, and restrained blue/orange/lilac overprint misregistration. All elements sit on the same frontal 2D plane, with a fixed head-on camera — no depth of field, no stereo camera, no 3D rotation, no spatial depth. 0–2s: On the off-white paper canvas, roughly a dozen small black, blue, orange, and lilac dots are scattered across the corners, edges, and near the center, at varying sizes with ample white space. The first blue dot appears abruptly with a short stamping sound and scales once; the surrounding dots wake in a chain reaction and jump one after another, each starting roughly 2 to 4 frames apart. They are not quietly breathing — they first dash a short distance in some direction, then abruptly brake, back off, and change direction, creating a clear sense of probing. Several dots enter at high speed from off-screen, striking the paper dots in frame and producing flat elastic rebound, trailing an extremely brief red/blue overprint ghost. The dots' motion paths covertly correspond to the Here letterform baseline: two columns on the left form the vertical motion channels of the H, dots in the middle trace the open arc of the e, and the right-hand cluster establishes the positional relationship of the r and the final e. 2–5s: The rhythm accelerates noticeably and the dots begin chasing and cross-swapping in pairs. The two left columns of dots shuttle up and down at high speed, and two black torn-paper vertical bars snap into place along their paths; a blue dot rockets left to right, and a short horizontal paper strip immediately snaps between the two verticals, briefly forming an H. The blue dot keeps accelerating into the middle, loops once to trace the open path of the e, and a blue circular paper piece snaps onto the inside of that path; another group of dots gathers into the vertical stroke of the r, then a short arc of paper flung in at speed connects as its shoulder; the trailing cluster forms the second e through rotation, compression, and rebound. The complete "Here" holds clearly for only about 0.2 seconds, with dots still racing through the interiors of the letters. Then a giant blue paper disc rapidly scales up from the edge of the frame to cover it entirely; the disc is instantly sliced into strips and arcs by four cuts, and the fragments are pulled forcefully in different directions, completing a high-speed collage transition. 5–8s: All the strokes and dots of the previous "Here" are flung to the left and right, while maintaining continuous relationships throughout. The two vertical paper bars of the H suddenly tilt and fold into the outer two diagonals of a W; two more connecting paper strips shoot in at high speed from the top and bottom of the frame; dots dash down along the four diagonal paths, undergoing pronounced collision, squash, and elastic rebound at the two valleys of the W, then rocketing back up, forming a continuously looping W motion skeleton. The circular paper piece from the previous e is pulled out of the layout, rotates a half turn, and adheres to the right of the W, its outer edge torn open to form an e. Black, blue, orange, and lilac dots each run along the four real strokes of the W at different speeds; after several rounds of interleaving they arrive at the turning points simultaneously on a strong beat and scale up in unison, and the complete "We" appears briefly and clearly. On the next beat the W is immediately squashed horizontally, then violently springs open like a coil; paper pieces fling off the letter edges, flip over, and re-adhere, sustaining a dense dynamic collage feel. 8–11s: The two left diagonals of the W keep converging at high speed, collide at the center, and spring upward to form the two diagonals of an A; a horizontal torn-paper strip carried over from Here sweeps in fast from the left and locks into the center of the A as its crossbar. The remaining paper strips on the right of the W are rapidly folded straight into the vertical stroke of the r; the open arc piece from the e rotates down from above, strikes the top of the r and bounces off to form the shoulder; another group of dots chases along an arc at speed and pastes up the final e. New dots join the Are structure continuously from all around, entering through three unclosed openings — the top of the A, the shoulder of the r, and the aperture of the e. Having entered, they do not stop, but continue running along the strokes at high speed, swapping positions and passing colors between them. After the complete "Are" appears briefly, the entire off-white page is rapidly covered by a giant black torn-paper sheet sweeping right to left, and the image instantly inverts to a black ground with off-white and blue/orange/lilac dots; within half a second another off-white sheet flips violently back up from below, creating a fast double page-turn, while the direction and position of the Are dots remain continuous throughout. 11–13s: Here, We, and Are appear simultaneously as a large magazine spread — Here on the left, We in the center, Are on the right — but the three words must not be displayed statically. All letters are constructed from dots and torn-paper strokes, with dots racing along the real strokes. The layout recomposes three times in succession: first, the three words offset vertically and snap back; second, the overall letter-spacing suddenly expands, with dots shuttling at speed through the gaps and drawing out connecting paper strips; third, the three words are magnetically snapped to a unified baseline. The open arc of the final e in Here rapidly unfurls toward the center and links into We; the rightmost diagonal of the W flings out to the right like a paper strip and folds into the A of Are. Large numbers of dots converge at high speed into the complete wordmark structure from different directions, but they do not merge into a closed center — instead they become letter terminals, stroke joints, and open entry points. The entire recomposition must feel like a magazine layout being rapidly swapped, torn open, pieced back, and calibrated on the beat, and must never become a slow alignment animation. 13–15s: After one strong overprint misregistration, all letterform structures rapidly complete a precise calibration, accurately forming "Here We Are." The final background stays off-white, the wordmark is centered, using an original modern geometric sans-serif design, with the body in deep ink black, signal blue emphasizing one connecting stroke or a local structure of "We," and coral orange and lilac retained only on a few letter terminals and surrounding dots. Once the wordmark forms it must not go completely still immediately: the surrounding dots first complete a fast half-lap chase around the outside of the wordmark, then abruptly snap to their corresponding letter terminals; all dots then pulse twice in succession at different rhythms, fully synchronizing on the second pulse on the same strong beat and briefly switching to signal blue. Finally an extremely thin 2D ink ripple spreads rapidly outward, producing one very slight overprint tremor as it passes the wordmark, after which everything locks immediately, and the complete "Here We