SSeedance 2.530s · 16:9Drag the clip sideways
A 15-second luxury perfume film built from @image1 (your bottle) and your notes (@image2 onward): each note gets its own macro shot, top notes first, the heart in the middle, the base at the end, then your bottle as the hero. Instrumental sound, no voice. Upload your perfume and its notes.
CHARACTER LOCK — HIGHEST PRIORITY: the person is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions. Preserve their gender presentation exactly as shown. Do not beautify, slim, de-age or restyle them. No other people in frame. Use they / their / them only. ATTACHED IMAGES ARE MANDATORY AND FINAL: the attached images outrank every word written below them. ASSET MAP: @image1 — the person, exactly as shown in the photo, unchanged and not regenerated throughout. Face, hair, skin, build, clothing and any accessories exactly as photographed. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. @image2 — the product, exactly as shown, unchanged and not regenerated — same shape, colour, materials, label/logo/patch and hardware in every shot it appears, even as the environment around it transforms. No invented branding or text added. @image2 itself never changes material or finish — only the light and setting around it transform. Ignore the grey studio pad around @image2 — that pad is not a set. SUBJECT LOCK — HIGHEST PRIORITY: the video opens on @image2 sitting on a plain, ordinary table, shot like a flat, dull phone photo — then transforms into a dramatic, cinematic hero setting around @image1 and @image2, with @image2 itself always staying exactly as shown. Do not beautify, restyle or redesign either subject. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a person. Only one instance of @image2 exists in the scene. PROP CONTINUITY: @image2 is visible from frame 1 onward in every single beat, without exception. CAUSALITY LOCK: the ordinary table and flat lighting are already cracking with bursting light and energy in frame 1 — not a slow build-up; @image1's hand is already reaching for @image2 as the transformation starts; each subsequent change in the environment is a direct, visible continuation of that first burst; @image1's lines are spoken in direct reaction to what is happening around them in that same beat, never disconnected from the visuals. PACING: the hook must land in the first full second — the ordinary photo is already erupting into transformation at 0s. Real, continuous motion for nearly the whole duration, calm only in the final hold. Never freeze on the reference still. SCENE: begins as a flat, ordinary table-top phone-photo look, then bursts into a dramatic, cinematic hero environment — warm dramatic light, rising energy, shallow depth of field, confident and triumphant tone throughout. Photoreal live-action. 0–1s: The ordinary photo of @image2 on a plain table is already cracking open with bursting light — @image1's hand already reaching into frame. DIALOGUE (@image1): “Watch this.” 1–4s: @image1 grabs @image2 as the flat, dull room dissolves and reforms into a dramatic, glowing environment around them. DIALOGUE (@image1): “From my camera roll—” 4–8s: The transformation completes — @image1 now stands in a striking, cinematic setting, @image2 held confidently, light rays and drifting embers of light around them. DIALOGUE (@image1): “—to this.” 8–12s: Camera orbits slowly around @image1 and @image2, showing the new environment in full from multiple angles. DIALOGUE (@image1): “Same product. Same photo.” 12–16s: @image1 raises @image2 and turns it naturally in the light, examining it with visible pride. DIALOGUE (@image1): “One upload. That’s it.” 16–20s: A burst of light sweeps across @image2's surface as @image1 holds it up. 20–24s: Camera pushes into a close macro shot of @image2, its real texture and materials catching the dramatic light exactly as shown. 24–27s: Camera pulls back to reveal @image1 standing tall with @image2 held high, the full cinematic environment visible behind them. DIALOGUE (@image1): “Try it with yours.” 27–30s: Motion slows into a confident final hold, @image1 and @image2 steady and triumphant, light settling calmly around them. ON-SCREEN CAPTIONS: word-timed captions of each DIALOGUE line above, clean sans-serif, bottom third, readable with sound off, shown only during beats that have a DIALOGUE line. AUDIO: @image1's own voice speaking the DIALOGUE lines, clear and confident, naturally lip-synced, plus a rising whoosh of energy at the transformation and ambient shimmer underneath, settling into quiet ambient tone at the final hold. Soft background music under the voice throughout, never louder than the voice. No BGM-only passages with no voice present until the final hold. No watermark, no subtitles beyond the word-timed captions described above.
ASSET MAP — every reference is bound here and cited by tag below: @image_1 — the fan. This is the person; gender, face, hair, skin and body proportions come entirely from @image_1 and not from this text. @image_1 is the identity source only - its clothing, pose, framing, background and lighting are ignored. @image_2 — the product the fan holds. Its shape, colour, material, label and every printed detail come entirely from @image_2 and not from this text. Keep @image_2 exactly as it is, at its true real-world size, held comfortably in one hand. ATTACHED IMAGES ARE MANDATORY AND FINAL: @image_1 is the person in every shot and @image_2 is the only product in the film. The attached images outrank every word written below them: wherever an image and this text disagree, follow the image. CHARACTER LOCK: @image_1 keeps the same facial identity, facial structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, distinctive marks, apparent age, build and body proportions in every shot, from the first frame to the last. The gender presentation of @image_1 stays exactly as shown in that reference; do not masculinize, feminize or otherwise reinterpret it. Do not beautify, slim, age or de-age @image_1. Use they / their / them only. PRODUCT LOCK: every letter visible on @image_2 is a letter that exists in its reference; never add lettering or a logo that is not on the reference, and never change its colours or proportions. A high-energy night baseball commercial under blazing stadium floodlights: a foul ball smashes straight into the camera, ricochets into the stands, and @image_1 catches it bare-handed while holding @image_2 perfectly still. Shot 1: a very short opening shot, under one second. Point-of-view of a camera at the edge of the field: a white baseball rockets out of the distance at blinding speed, grows from a dot to filling the whole frame in an instant, and slams straight into the lens of a plain black camera with no text on it; the image jolts violently and a spiderweb crack flashes across the glass. <whoosh> <hard glass impact> <deep sub-bass boom> Shot 2: extreme macro, ultra slow motion: the spinning baseball fills the frame as it ricochets away, red stitches whipping past, leather texture razor sharp, tiny glass shards and dust trailing behind it, floodlight glinting on the seams. <slow-motion rumble> Shot 3: the ball arcs high up into the stands, the camera whipping after it toward the front row. <crowd gasp> Shot 4: @image_1 sits in the front row, eyes snapping up toward the incoming ball, @image_2 held at chest height in the right hand, its front facing the camera, the left hand empty. Shot 5: ultra slow motion, low angle. @image_1 rises from the seat; the empty left arm lifts high overhead, the left palm opens, and the left hand closes around the ball with a clean slap. The right hand stays down at chest height holding @image_2 level and still, never raised; popcorn hangs in the air behind. <leather slap> Shot 6: back to real speed, straight after the catch the left hand keeps the ball raised high overhead and never lowers it. The section around @image_1 erupts, neighbours grab @image_1's shoulders, @image_1 shouts with joy while the right hand holds @image_2 at chest height, fireworks bursting over the stadium. <stadium roar> <firework bursts> Shot 7: extreme macro close-up of @image_2 only: floodlight and the glow of the fireworks glide slowly across its surface as moving sparks, every edge, material and printed detail of @image_2 razor sharp and clearly readable. <deep bass hit> Shot 8: @image_1 lifts @image_2 in the right hand toward the camera; fast push-in until @image_2 fills the centre of the frame, @image_1 grinning soft-focus behind. Wardrobe (from THIS TEXT, never from @image_1): @image_1 wears plain, unbranded casual game-night clothes that suit them, no hat, hair uncovered. Every garment is cut unisex and keeps its design, cut, proportions and colour whoever appears in @image_1; only its drape adapts to the body. Everyone in the crowd wears plain black, grey or white hoodies and T-shirts; nobody in the crowd wears a cap, a jersey, or any sports gear. The crowd behind @image_1 is always soft and out of focus. Location: all stadium walls, padding, boards and screens are plain solid navy or plain glowing colour with no text, symbols or logos anywhere. Cinematography: hyper-real sports commercial, anamorphic lens, shallow depth of field, hard white floodlight with lens flares, deep night sky, rich contrast, crisp detail, punchy speed ramps. (Loud driving electronic trap beat that leads the whole soundtrack: one huge bass hit exactly on the ball smashing into the lens, a rising riser through the catch, a heavy bass drop as the crowd erupts, a hard hit on the product macro, a final impact hit on the last shot.) Hands: hands stay anatomically correct with five fingers on each hand wherever a hand is visible. @image_2 is only ever in the right hand and the ball is only ever in the left hand; the ball and @image_2 never touch. Only one fan, one ball and one @image_2 exist in this video; do not generate duplicate fans or a twin effect. Keep it subtitle-free, avoid generating any text or subtitles, keep @image_2's own label and logo exactly as shown and add no other logo, do not generate a watermark, no team crests or lettering on any clothing.
VICTORIAN LONDON DETECTIVE — THE CASE OF THE STOLEN DIAMOND. A moody, atmospheric 1890s mystery told as one clear, logical story in seven scenes, at natural realtime speed. Every clue the detective finds leads directly to the next scene, every action is shown slowly and completely, every object is picked up and put down by hand, and every object stays exactly where it was placed. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a tense, elegant Victorian mystery score with pizzicato strings, a ticking-clock rhythm, a lone violin and a dramatic orchestral reveal at the climax; original, instrumental only. 2. SFX — quiet, under the music: horse hooves on wet cobblestones, rain, the ticking of a watch, footsteps on wooden floors, a gasp from a small crowd. 3. VOICE — none; @image1 stays silent the whole clip. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. AUDIO QUALITY: the whole soundtrack is one clean, crisp, studio-mastered stereo mix — the music plays continuously and smoothly from the first frame to the final hit, every pause in the action is filled by the music itself, and every sound effect is clear, natural and matches a visible action on screen. TWO references with separate jobs: @image1 is the detective — @image1 keeps the exact same face and features as the reference the whole time, only what is present in the reference. Preserve their gender presentation exactly as shown. Use they / their / them only. @image2 is THE WATCH — take only the watch from @image2 and keep it exactly as the reference whenever it appears: its exact case, colour, dial, hands and every detail exactly as shown, nothing added or redesigned. The @image2 watch lives in @image1's left waistcoat pocket and comes out only when @image1 draws it out by hand. SKIN AND LIGHT: every person in every scene keeps a natural, healthy skin tone — @image1 keeps their exact skin tone as shown in @image1 — lit by the warm amber gaslight and candlelight. WARDROBE — the same in every scene: @image1 wears a Victorian detective's outfit styled to suit their gender presentation as shown in @image1 — brown tweed with a waistcoat, a white shirt, a long charcoal wool overcoat and a black bowler hat. THE CULPRIT — a separate person: a tall, elegant guest of the mansion in a black evening tailcoat with a short black cape lined in bright scarlet silk worn over the shoulders, the face always fully visible and uncovered, with neat dark hair and a thin, knowing smile. The scarlet lining of the cape has one small torn corner. THE DIAMOND — a large, brilliant-cut clear diamond the size of a walnut, sparkling in any light. BEATS (30s): 0–4s: THE SCENE OF THE CRIME — night in a grand Victorian mansion study lit by gas lamps. A tall glass display case stands on a mahogany pedestal; its front glass panel is completely smashed open, a large jagged hole the size of a person's head in the front, glass shards scattered across the Persian carpet, and inside it an empty royal-blue velvet cushion with a round dent where the diamond rested. @image1 walks in behind a nervous grey-haired butler, stops in front of the case and studies the empty cushion through the broken glass. Camera: wide from the doorway, then a slow push-in on the empty velvet cushion. 4–8s: THE BOOT PRINT — beside the pedestal, on the cream carpet, lies a single clear muddy print of a leather dress shoe, the heel and sole sharply outlined in dark, wet river mud with tiny flecks of green weed. @image1 kneels on one knee, takes a brass magnifying glass from the overcoat pocket and studies the shoe print closely, then touches the mud with a fingertip and rubs it between the fingers. Camera: close-up through the magnifying glass on the shoe print, then on @image1's thoughtful face. 8–12s: THE SCARLET THREAD — @image1 rises and walks to the tall window beside the display case; the window is open, and caught on its brass latch is a small torn triangle of bright scarlet silk. @image1 pulls it free between finger and thumb and holds it up against the gaslight, turning it slowly. Camera: close-up on the fingers and the scarlet silk, the gaslight glowing behind it. 12–15s: THE WATCH — @image1 draws the @image2 watch out of the waistcoat pocket, holds it in the palm with its face exactly as @image2 shows it, reads the time — a quarter past eleven — and slips the @image2 watch back into the waistcoat pocket. Camera: tight close-up on the @image2 watch in the palm, then on the narrowed eyes. 15–19s: THE GUESTS — in the candlelit drawing room the evening's guests stand in a loose line: a guest in a lilac silk evening outfit, an old white-haired colonel, and the tall guest in the black tailcoat and scarlet-lined cape. @image1 walks slowly along the line, looking down at each guest's shoes; the tall guest's polished shoes are caked with the same wet river mud and green weed. Camera: low tracking shot along the line of shoes, then rising to the tall guest's face. 19–24s: THE PROOF — @image1 stops in front of the tall guest, lifts the corner of the scarlet-lined cape with one hand and holds the torn scarlet silk triangle against it; it fits the torn corner exactly. Camera: close-up on the scarlet silk fitting the torn corner of the cape, then a medium shot of both faces. 24–28s: THE DIAMOND — @image1 reaches into the inner breast pocket of the tall guest's tailcoat and slowly draws out the large brilliant diamond, holding it up between finger and thumb in the candlelight so it blazes with sparkling light. The other guests gasp, and two police constables in tall helmets step in and take the tall guest gently by both arms, the guest's warm-toned face clearly visible, lips pressed together in defeat. Camera: close-up on the blazing diamond, then a wide of the room. 28–30s: THE END — @image1 places the diamond back on the royal-blue velvet cushion in the broken display case, draws out the @image2 watch once more, glances at it with a small satisfied smile, tips the bowler hat and walks out. Camera: medium on the diamond back on the cushion, then a wide of @image1 leaving, holding to the last frame. LOOK: ultra-photorealistic Victorian period drama, warm amber gaslight and candlelight against deep shadows, rich mahogany and velvet, natural motion blur, fine film grain, clean full frame.
VERTICAL 9:16 15 seconds. Seedance 2.5. CINEMATIC luxury-product commercial. Douyin-style transformation. The RESULT is from @image2. Photoreal. Dramatic lighting, high-end framing. Immersive whoosh, flash, wind, score. No on-screen captions. No spoken brand names. RIGHT-SIDE UP from frame 0. NEVER a studio still as frame 0. CHARACTER LOCK: the person is exactly @image1. Preserve face, hair, skin, age, build, gender presentation. Use they / their / them only. @image1 = FACE only. Photocopy this person. Looks: (a) BEFORE — bare face, messy hair, oversized T-shirt, sleepy, casual room. (b) AFTER — same person, glamorous, styled, elegant fashion, black studio. The after-look is the result of @image2. @image2 = THIS product. Photocopy it. NEVER a different product. They present @image2 to camera before the swipe. After the cut, @image2 is the hero still-life. One person only as hero. 0.00–3.50 BEFORE. Bare face, messy hair, oversized tee, sleepy, casual room. @image2 on the vanity — they hold it up to camera. Slow push-in. They become confident and FAST sideways palm-swipe the lens. Whip-pan with their hand; palm covers the frame. 3.50 HARD CUT + WHITE FLASH + WHOOSH + SHAKE. Instant hidden transformation. No morph. SNAP ZOOM. 3.50–7.00 AFTER. Black studio. Same person as @image1, glamorous. Fast head turn, intense eye contact. @image2 sharp in the foreground. 7.00–15.00 EDITORIAL. Rapid FLASH → HARD CUT → NEW POSE. Hair sweep (SLOW-MO). 3/4 turn. Hand near jaw. Low-angle. @image2 inserts between poses. Final hero: them + @image2. Huge white photography flash + brief freeze. Same person as @image1. Same product as @image2.
CHARACTER LOCK — HIGHEST PRIORITY: the person is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions. Preserve their gender presentation exactly as shown. Do not beautify, slim, de-age or restyle them. No other people in frame. Use they / their / them only. ATTACHED IMAGES ARE MANDATORY AND FINAL: the attached images outrank every word written below them. ASSET MAP: @image1 — the person, exactly as shown in the photo, unchanged and not regenerated throughout. Face, hair, skin, build, clothing and any accessories exactly as photographed. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. @image2 — the product, exactly as shown, unchanged and not regenerated — same shape, colour, materials, label/logo/patch and hardware in every shot it appears, even as the environment around it transforms. No invented branding or text added. @image2 itself never changes material or finish — only the light and setting around it transform. Ignore the grey studio pad around @image2 — that pad is not a set. SUBJECT LOCK — HIGHEST PRIORITY: the clip NEVER opens as a still photo. Frame 0 is already mid-burst: the dull table under @image2 is already cracking, light already exploding, @image1's hand already snatching @image2, camera already moving. Then the room rips into a cinematic hero space around @image1 and @image2. @image2 itself always stays exactly as shown. ANTI-STATIC LOCK: no posed standstill, no mannequin hold, no slow orbit around a still person, no lingering product-turn, no long macro, no end freeze longer than half a second. The camera never rests. @image1 never stops moving. SPEECH LOCK: @image1 speaks on camera with visible lip sync. They say exactly this line, once, clearly, in English, while walking and raising @image2 (around 8–12s): "This is the one." No other words. Mouth motion matches those four words. Not a voiceover, not off-screen. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a person. Only one instance of @image2 exists in the scene. PROP CONTINUITY: @image2 is visible from frame 1 onward in every single beat, without exception. CAUSALITY LOCK: the ordinary table is already detonating in frame 1; @image1's grab is already happening; every later change is a continuation of that first explosion. PACING: 15 seconds total. Hook at 0.00s. Continuous motion. Only the last half-second settles. SCENE: dull table-top already tearing apart, then a dramatic cinematic hero environment — warm hard light, flying debris, shallow depth of field. Photoreal live-action. Handheld + tracking. 0–1s: HOOK. Smash-in on @image2 as the table already cracks and light blasts through it — @image1's hand already grabbing @image2. No speech yet. 1–4s: @image1 yanks @image2 off the table and spins as the dull room rips away; camera tracks the spin. 4–8s: @image1 walks hard toward camera through the new glowing space, @image2 in hand; camera dollies backward. 8–12s: They raise @image2 while still walking and speak, lips matching: "This is the one." 12–15s: Last burst of light, they take one more step into camera, half-second settle. No extra line. AUDIO: native speech plus foley. Explosive whoosh at 0s, rushing air, debris, footsteps, fabric, a sharp grab. At 8–12s, @image1's own voice clearly says "This is the one." No music, no BGM, no voiceover, no narration, no second speaker. No on-screen text, no watermark, no subtitles.
THE LEGEND OF BLACKWOOD MANOR — A GHOST HUNTER'S NIGHT. A spooky-funny, heart-warming ghost story told as one clear story from start to finish in seven scenes, at natural realtime speed, with realistic physics for everything solid and a soft, floating, weightless glide for the ghost. A real ghost has lived in Blackwood Manor for two hundred years, frightening every visitor away, until @image1, a ghost hunter, discovers what the ghost has really been missing all this time. AUDIO MIX (ranked, three separate layers): 1. MUSIC — loudest, dominant from the very first frame: a playful, spooky cinematic score with a creepy harpsichord, pizzicato strings, a theremin and bassoon for comedy, swelling into a warm, glowing, heavenly orchestral finale with a soaring wordless choir pad; original, instrumental only. 2. SFX — quiet, under the music: wind howling, creaking floorboards, rattling chains, a loud comical ghostly stomach rumble, the crinkle of a chocolate wrapper, a crunchy bite, a soft heavenly chime. 3. VOICE — none; @image1 stays silent the whole clip. Audio suppressor: no voiceover narration, no dialogue, no spoken words, no singing vocals, no lyrics, no library audio bed, no muffled murmuring, no subtitles. The music resolves on a final hit and silence holds the final 0.5s. AUDIO QUALITY: the whole soundtrack is one clean, crisp, studio-mastered stereo mix — the music plays continuously and smoothly from the first frame to the final hit, every pause in the action is filled by the music itself, and every sound effect is clear, natural and matches a visible action on screen. TWO references with separate jobs: @image1 is the ghost hunter — @image1 keeps the exact same face and features as the reference the whole time, only what is present in the reference. Preserve their gender presentation exactly as shown. Use they / their / them only. @image2 is THE CHOCOLATE BAR — take only the chocolate bar from @image2 and keep it exactly as the reference whenever it appears: its exact wrapper shape, colours, logo and lettering. The @image2 chocolate bar sits in the left chest pocket of @image1's field jacket until @image1 takes it out by hand. WARDROBE — the same in every scene: @image1 wears a dark olive field jacket with a chest pocket, black jeans, leather boots, a small canvas backpack, and holds a plain black torch in the right hand. THE GHOST — a tall, thin Victorian aristocrat ghost with a clearly different face from @image1: a long pale face, sunken sad eyes, droopy grey eyebrows, a top hat and a tattered tailcoat with a rusty chain around one ankle. The ghost is translucent, glowing a cold pale blue, the walls visible through the body, and floats a few centimetres above the floor. BEATS (30s): 0–4s: THE MANOR — at midnight under a full moon, @image1 pushes open the creaking front door of the dusty manor and steps into the grand hall, sweeping the torch beam across sheet-covered furniture and cobwebbed portraits. Camera: wide from inside the hall, the door swinging open. 4–8s: THE GHOST APPEARS — the translucent ghost rises up through the floorboards right in front of @image1, spreads its arms wide, rattles its chain and puts on its most terrifying face. @image1 jumps back, the torch beam shaking wildly, eyes wide. Camera: low behind @image1, then a close-up on @image1's shocked face. 8–12s: THE RUMBLE — just as the ghost leans in to scare again, a huge, loud, comical rumble comes from the ghost's see-through belly. The ghost freezes, looks down at its own stomach, then up at @image1 with a sad, embarrassed, hungry face, its eyebrows drooping. @image1 tilts the head, curious. Camera: medium two-shot, then a close-up on the ghost's glowing belly. 12–16s: THE GIFT — @image1 reaches into the chest pocket with the free left hand, takes out the sealed @image2 chocolate bar, still fully closed in its wrapper, and holds it out to the ghost on the open left palm, logo facing the camera, the torch still in the right hand. The ghost leans down, sniffs the air, and its eyes grow huge and sparkling. Camera: close-up on the sealed @image2 bar on @image1's palm, then the ghost's amazed face. 16–22s: THE FEAST — the ghost takes the sealed @image2 bar from @image1's palm and opens it with BOTH of its own translucent hands: its left hand holds the bottom of the bar, its right hand tears the top of the wrapper open with a loud crinkle and peels the wrapper down, showing the bare chocolate. Then the ghost EATS the chocolate, bite after bite: it brings the bar to its mouth and bites off the top, chewing with puffed cheeks, the bitten piece glowing golden as it slides down through its see-through body into the belly; it takes a second big bite, chewing happily, the bar now half gone; then a third bite, and the bar is completely eaten, only the empty, flattened wrapper left in its fingers. The ghost crumples the empty wrapper into a small ball and tucks it into its tailcoat pocket, licks its fingers and pats its glowing full belly, eyes closed in pure bliss, mouth lifting into a huge happy smile. @image1 laughs. Camera: tight close-up on the two hands tearing the wrapper, then close on the mouth biting and chewing each bite as the bar gets shorter, the glowing chocolate visible inside the see-through belly. 22–26s: THE GLOW — the rusty chain around the ghost's ankle falls open and drops to the floor with a clank. Warm golden light starts glowing from the ghost's full belly and spreads through its whole body, turning the cold blue to bright gold. The ghost tips its top hat to @image1 with deep gratitude. Camera: slow push-in as the golden light fills the hall. 26–30s: TO HEAVEN — the golden glowing ghost floats slowly upward, waving down at @image1, rising through a shaft of warm heavenly light that opens in the ceiling, up past the chandelier and through the roof into a starry sky full of soft golden clouds. @image1 stands below with the torch in the right hand and the empty left hand raised high, waving goodbye with an open palm, and smiles. Camera: low from behind @image1 looking up into the beam of light, holding to the last frame. LOOK: ultra-photorealistic cinematic fantasy-comedy grade, deep moonlit blues and dusty shadows turning into warm heavenly gold, soft volumetric light, natural motion blur, fine film grain, clean full frame.
CHARACTER LOCK — HIGHEST PRIORITY: the person is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions. Preserve their gender presentation exactly as shown. Do not beautify, slim, de-age or restyle them. No other people in frame. Use they / their / them only. ATTACHED IMAGES ARE MANDATORY AND FINAL: the attached images outrank every word written below them. ASSET MAP: @image1 — the person, exactly as shown in the photo, unchanged and not regenerated throughout. Face, hair, skin, build, clothing and any accessories exactly as photographed. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. @image2 — the product, exactly as shown, unchanged and not regenerated — same shape, colour, materials, label/logo/patch and hardware in every shot it appears. No invented branding or text added. Ignore the grey studio pad around @image2 — that pad is not a set. SUBJECT LOCK — HIGHEST PRIORITY: @image1 stands at a bright kitchen counter, speaking directly to camera with lips naturally in sync with every spoken line. @image2 is held in @image1's hand or resting right beside them, clearly visible in every single moment of the video until the end card. Do not beautify, restyle or redesign either subject. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned. SINGLE SUBJECT: only @image1 appears as a person — no crew, no other people, no second hand. Only one instance of @image2 exists. CAUSALITY LOCK: the camera’s slow pull-back at 12–16s is a smooth continuous move that reveals the empty room — no tripod, no lights, no crew, nobody else present; @image1 reacts to what the pull-back shows with a small amused look. Nothing appears or disappears without a visible cause. PACING: the video opens already in motion, mid-sentence energy, no still establishing shot. Real continuous motion throughout (gestures, small movements, steam rising from a cup, soft light shifting). Never freeze on the reference still. SCENE: a bright kitchen counter, soft morning window light, shallow depth of field, realistic handheld-style framing that stays steady. Photoreal live-action. 0–2s: Close, steady framing. @image1 holds @image2 up beside their face and speaks straight to camera. DIALOGUE (@image1): “I hate being on camera.” 2–5s: A small knowing smile, @image2 turned slightly toward the lens. DIALOGUE (@image1): “So I didn’t film this.” 5–9s: @image1 lifts @image2 a little higher and counts on the free hand. DIALOGUE (@image1): “One selfie. One product photo.” 9–12s: @image1 gestures with the free hand, relaxed and confident. DIALOGUE (@image1): “That’s everything I gave it.” 12–16s: The camera pulls back smoothly and reveals the whole kitchen: empty, with no camera crew, no lights, no tripod. @image1 glances around, then back to camera, amused. DIALOGUE (@image1): “Who’s filming? Nobody.” 16–21s: The camera settles back in close. @image1 steps toward the lens and offers @image2 toward camera. DIALOGUE (@image1): “Your face. Your product. Your ad.” 21–27s: @image1 sets @image2 down, then picks it up again as if starting over, confident and light. DIALOGUE (@image1): “Swap in your own and run it again.” 27–30s: Cut to a clean end card on a plain dark background, centred text only. ON-SCREEN TEXT (end card only): “Sign up at openclips.ai. No card.” ON-SCREEN CAPTIONS (0–27s): word-timed captions of every DIALOGUE line, bold clean sans-serif, large, bottom third, one key word per line highlighted in a bright accent colour, readable with the sound off, no captions during the end card. AUDIO: @image1's own natural voice speaking the DIALOGUE lines, clear and close, lips in sync, with soft kitchen room tone (a quiet cup clink, a faint hum). Light, warm background music under the voice throughout, never louder than the voice, fading out during the end card. No sudden volume jumps, no crackle. No watermark. No logos anywhere except the plain end-card text and the product’s own marks from @image2.
subject_definitions: <Subject 1> is the dancer whose appearance comes from <Picture 1> and whose dance moves come from <Video 1>. <Picture 1> is the person: gender, face, hair, skin, apparent age and body proportions come entirely from <Picture 1> and not from this text or from <Video 1>. <Picture 1> defines identity and clothing - its pose, framing, background and lighting are ignored. <Video 1> gives only the dance moves, their rhythm, the locked-off full-length framing and the burning kitchen - the dancer seen in <Video 1>, their face, hair, body and clothing are never copied. <Subject 1> wears their own clothes as seen in <Picture 1>, continued naturally to full length. <Subject 2> is the home kitchen that is fully on fire, whose layout, flames and smoke follow <Video 1>. <Audio 1> is the complete soundtrack of the target video, a song with sung vocals. summary: [reference generation + audio reuse] The target video is a 15-second vertical Instagram Reel in one continuous locked-off shot: <Subject 1> performs the joyful dance from <Video 1> inside <Subject 2>, already burning wall to wall from the first frame, beaming at the lens, with <Audio 1> as the soundtrack; near the end giant white letters reading "BE OK" appear on the burning wall behind <Subject 1>. retention_analysis: <Subject 1> (appears in [Shot 1]): fully_preserved - the person in every frame is exactly the person in <Picture 1>: facial identity and structure, eye shape and spacing, eyebrows, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, distinctive marks, apparent age, gender presentation, build and body proportions, unchanged from the first frame to the last; not beautified, slimmed, de-aged or restyled; wearing their own clothes from <Picture 1>. <Subject 2> (appears in [Shot 1]): partially_preserved - the same burning kitchen, flames and black smoke as in <Video 1>. <Video 1> (dance moves and framing): attribute_transfer - the dance moves, their timing and the full-length locked-off framing are transferred to <Subject 1>; nothing of the dancer in <Video 1> is kept. <Audio 1>: fully_copy - <Audio 1> plays unchanged from 0 seconds to the end as the complete soundtrack. detailed_description: The target video is live-action and photoreal, lit by the warm orange light of the fire, with crisp detail, real flames, rolling black smoke and fine grain. Deadpan comedy: <Subject 1> is radiantly happy and completely unbothered, as if nothing unusual is happening. No other people appear. [Shot 1] A full-length vertical shot frames <Subject 1> from the feet to above the head in the middle of <Subject 2>; every upper cabinet, the open shelves, the curtains, the cooker hood and the ceiling are already burning in tall bright orange flames from the very first frame. The camera holds a static shot for the whole video. The shot opens mid-dance: <Subject 1> skips toward the lens with a happy kick, their clothes swinging with the move, as a burning cabinet door falls behind in a burst of sparks. <Subject 1> then dances the moves of <Video 1> with light, carefree, joyful energy on every beat of <Audio 1>: skipping steps, a spin, playful leg kicks, hands clasped under the chin, arms thrown up in delight. Around the ninth second the fire behind flares even brighter while <Subject 1> spins again and throws both arms up toward the ceiling. From the twelfth second <Subject 1> stands in the centre of the frame facing the lens, arms open wide, holding a radiant smile and blinking once. At the same moment giant bold white block letters reading "BE OK" appear on the burning wall behind <Subject 1>, filling the width of the frame at chest-to-head height; the body and open arms of <Subject 1> stay in front of the letters and partly cover them, and the flames keep moving around them until the last frame. <Subject 1>'s hands always have five fingers. <Subject 1>'s lips stay closed in a smile or open in a silent laugh and never sing along. <Subject 1>'s face and clothes stay the same from the first frame to the last. overall_soundscape: The roar and steady crackle of the fire through the whole shot, footsteps tapping on the kitchen floor, the crash of the falling cabinet door in the first second, a deep whoosh as the fire flares around the ninth second, all kept quiet under <Audio 1>. non_diegetic_music: <Audio 1> as given, from the first frame to the last, at one even level.
CHARACTER LOCK — HIGHEST PRIORITY: the person is exactly the person in @image1. Preserve their facial identity, facial structure, eye shape and spacing, nose, mouth, exact skin tone, hairline, hairstyle, hair length, texture and colour, freckles and distinctive marks, apparent age, build and body proportions. Preserve their gender presentation exactly as shown. Do not beautify, slim, de-age or restyle them. No other people in frame. Use they / their / them only. ATTACHED IMAGES ARE MANDATORY AND FINAL: the attached images outrank every word written below them. ASSET MAP: @image1 — the person, exactly as shown in the photo, unchanged and not regenerated throughout. Face, hair, skin, build, clothing and any accessories exactly as photographed. Ignore the grey studio pad around @image1 — that pad is not a set. Never freeze on the attached still as frame 0. Never show a checkerboard or transparency grid. @image2 — the product, exactly as shown, unchanged and not regenerated — same shape, colour, materials, label/logo/patch and hardware in every shot it appears. No invented branding or text added. Ignore the grey studio pad around @image2 — that pad is not a set. Bring nothing from the background of the @image2 photo into the living room — only the product itself. SUBJECT LOCK — HIGHEST PRIORITY: @image1 sits on a couch in a dim living room, a TV glowing off-frame in front of them (the screen itself is never shown — only its light). @image2 rests on a coffee table in front of @image1, clearly visible from the very first frame where it appears. Do not beautify, restyle or redesign either subject. HANDS: hands are always anatomically correct with five fingers each, naturally proportioned, in every shot they appear. SINGLE SUBJECT: only @image1 appears as a person — no other people enter frame. Only one instance of @image2 exists in the scene. PROP CONTINUITY: the dim room, the off-frame TV’s glow, @image1 and @image2 on the coffee table are all established in the opening beat — nothing new appears later unexplained except the party poppers, which are caused by @image1's own celebration (see CAUSALITY LOCK). CAUSALITY LOCK: the TV’s off-frame light bursts into an intense flash at the exact instant of frame 1 — this is the game-winning moment, never shown on screen, only felt through the light and @image1's reaction; @image1 is already leaping up from the couch in that same instant; the flashing light sweeps across the room in a celebratory rhythm, and as it passes over @image2 it leaves a warm passing highlight on its surface, its own colours unchanged, like a trophy catching a spotlight — a direct visible effect of the light passing over it, not an unexplained glow; @image1 then grabs @image2 and raises it overhead; the party poppers going off are a direct, visible response to @image1 raising @image2 overhead, never an unexplained trigger. PACING: the hook must land in the first full second — the TV’s flash and @image1's leap happen together at 0s, not building up to it. Real, continuous motion for the whole duration, calm only in the final macro beat. Never freeze on the reference still. SCENE: a dim, cosy living room at night, the only light source the off-frame TV’s pulsing glow, shallow depth of field, warm cinematic realism. Photoreal live-action. 0–1s: The off-frame TV bursts into an intense flash of light — @image1 is already leaping up from the couch in that same instant, arms starting to rise. 1–3s: The light keeps pulsing in a celebratory rhythm across the room, @image1 mid-celebration, @image2 still clearly visible on the coffee table. 3–6s: As the pulsing light sweeps over @image2, a warm highlight ripples across its surface, its own colours unchanged, like a trophy catching a spotlight. 6–9s: @image1 reaches down, grabs @image2, and raises it overhead in triumph. 9–12s: Party poppers on the side tables go off in response, streamers and confetti bursting into the air around @image1. 12–15s: Camera pushes into a close macro shot of @image2 held aloft, confetti drifting softly past, the TV’s glow still pulsing warmly out of focus behind. AUDIO: natural home-celebration foley only — the TV’s sudden volume swell (muffled, no actual commentary words audible), @image1's gasp and cheer, the pop and hiss of party poppers, falling confetti rustle, fading to a warm quiet hum on the macro shot. No music, no BGM, no voiceover, no dialogue, no narration. No on-screen text, no watermark, no subtitles.
Zombie Fight — Everyone Wants a Sip (Your Drink)Seedance 2.5 · 16:9 · 30sRecreate workflow