Crowd / Group

A cinematic Japanese slice-of-life scene. Two young Japanese women (aged 20 and 22) are sitting together and chatting happily. They are close friends having a relaxed conversation. Their lips move naturally with synchronized dialogue (accurate lip sync). Scene: a quiet residential street in Japan in the late afternoon. Warm golden sunlight, gentle breeze, peaceful everyday: cinematic Japanese film style composition, medium two-shot, natural framing, slightly handheld camera, slow cinematic: soft natural sunlight, realistic skin tones, subtle film color grading. Visual style: Japanese indie film aesthetics, shallow depth of field, soft background blur, realistic acting. Dialogue (spoken in Japanese, natural lip sync): Woman 1: 'It's really warm today. Doesn't it feel like spring?' Woman 2: 'Yeah, it feels good to talk outside on a day like this.' Woman 1: 'How've you been lately? Is university busy?' Woman 2: 'A bit busy, but I feel better talking like this.' They smile, laugh lightly, and continue their casual conversation. Ultra realistic, cinematic storytelling, natural acting, 4K film look.

A cinematic 10-second scene on the red rocky surface of Mars at golden hour. A visionary space pioneer wearing a black baseball cap that reads, “I would like to die on Mars. Just not on impact,” and matching black Mars socks with colorful planet designs visible above rugged boots, slowly walks across the landscape. He pauses, looking toward the distant horizon with quiet curiosity, as if wondering whether someone else who shares his dream has already arrived. In the distance, a woman wearing the same distinctive black cap stands on a ridge, their shared cap becoming a silent symbol of hope, exploration, and human connection. Soft Martian wind, realistic dust, cinematic lighting, ultra-realistic, highly detailed, emotionally uplifting, medium shot with both the hat and socks clearly visible.

animate this hokey players statues to show them like entering the ice and prepare for a photo shoot

The princess continues walking gracefully forward along the rose path, her sheer embroidered gown and long veil flowing elegantly with her steps as golden light shines. She approaches the handsome king standing at the rose-covered castle archway…

Ultra-realistic original dark fantasy war sequence on a vast volcanic plain at dusk. An army of terrifying beast riders mounted on armored reptilian and wolf-like creatures charges toward a disciplined line of elite spearmen. 0–4s: extreme wide shot showing hundreds of mounted creatures racing across ash and burning grass, dust clouds and embers exploding behind them. 4–9s: low-angle tracking beside the charge, claws pounding the ground, riders roaring, capes and armor shaking, camera weaving between galloping monsters with intense speed. 9–15s: full-force impact as the beast riders crash into the shield wall, spears splinter, bodies recoil, creatures leap through smoke, the formation bends under enormous pressure. Epic scale, brutal energy, ultra dynamic camera, photorealistic textures, cinematic dust, ash, sparks, deep orange firelight against black storm clouds.

[Overall Style] Cinematic photorealistic, restrained supernatural twist comedy, combining elegant immortal time spells, precise visual misdirection, realistic physics of static objects, and subtle character micro-performances. The humor comes from the accidental exposure of a character's true nature. Designed for Seedance 2.0 multi-modal reference control and unified audio-visual generation. [Characters] Character A | Sword Immortal Sister: Use @Picture 1 as the strict identity and clothing reference. A 25–30-year-old East Asian female sword immortal, maintaining the same face, sharp dark eyes, black long hair, jade hairpin, tall slender figure, white embroidered silk Hanfu, translucent layered wide sleeves, silver jewelry, and white cloth boots. Character B | Bicycle Sister: Use @Picture 2 as the strict reference. A 25–30-year-old East Asian female on a bicycle, maintaining the same face, ponytail, body proportions, clothing, accessories, bicycle, and bubble tea cup. [Shot 1 | 0-5s | Low-angle Panorama Tracking] 16:9 widescreen, a bright modern pedestrian plaza after rain. Wet ground reflections, coffee tables, trees, glass storefronts, and leaves blowing in the wind are clearly visible. The same bicycle girl enters the frame, her front wheel slips slightly, and she loses balance. The bubble tea cup, black pearls, mobile phone, and scattered leaves fly into the air. The same sword immortal turns and whispers: 'Freeze.' Time instantly stops; all objects in the air, tea droplets, and pearls hover precisely. Urban sounds vanish, replaced by a low, clear magical resonance. [Shot 2 | 5-10s | Continuous Orbiting Medium Shot] The same sword immortal in white silk Hanfu walks calmly into the static scene. She straightens the bicycle, puts the phone back in the girl's hand, catches all suspended pearls with the tea cup, and gently adjusts the girl's center of gravity. Before leaving, she looks around, secretly takes a quick sip of the bubble tea, then immediately resumes her cool posture with hands behind her back. [Shot 3 | 10-15s | Close-up into Two-person Shot] The sword immortal snaps her fingers, and time resumes seamlessly. Character B lands safely with a shocked expression. She checks the suddenly lighter tea cup, notices a black pearl on the immortal's lip, and asks suspiciously: 'Sister, did you freeze time just for a sip?' Extreme close-up: The immortal silently swallows the tea, wipes the pearl away, and answers calmly: 'Saving people, by the way.' The frame freezes on the girl's suspicious gaze and the immortal's forced calm. [Technical Requirements] Strict 15-second total duration, three clean shots, consistent object trajectories, stable character identities and clothing, accurate Mandarin lip-syncing, precise audio transition for time freezing, and realistic physics for liquids, pearls, hair, silk, bicycles, and reflections. No subtitles. [Negative Prompts] blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props

Act 1: [Reference Lock] Refer to Image 1 hf_20260730_044943_833909d3-0181-4d86-b4ad-8fdd91945fbd for Landlady #1 wearing hf_20260730_045132_8f945ce1-0e33-4e9c-86c6-5b5bdb0b0185. Legs and feet proportions, thin sock texture, sofa sitting posture, and black high heel position are highest priority; maintain pre-set face, hairstyle, makeup, and clothing of #1. Reference video provides the small restaurant space, appearance, body type, clothing, and life-like performance rhythm of Man #2 and Diner #3. #1, #2, and #3 are all adults; no face-swapping, duplication, merging, or identity swapping. [Overall Setting] Realistic small restaurant life comedy. Landlady #1 sits on the sofa in the left-rear resting area browsing her phone, presenting a restrained, natural 'Siren Pose': body leaning against the sofa, one shoulder slightly lower, waist and back forming a natural S-curve; legs elegantly crossed and extended slightly forward with toes relaxed; head tilted gently, long black hair falling naturally on one shoulder. The posture has a quiet, lazy, attractive quality but does not actively tease diners, lick lips, wink, or intentionally writhe. Man #2 sits at the first old green dining table in the front right eating noodles. Diner #3 sits at another independent dining table in the back right; there is a clear aisle between the two tables, and they never sit at the same table. Man #2 stares momentarily at Landlady #1 while eating, then calls her to get a bottle of orange drink. Diner #3 is already eating at the other table from his first appearance to establish spatial positioning for the later twist. [Realistic Filming] Unprocessed iPhone handheld realistic video. 9:16 vertical screen, 1080x1920, 30fps, 26-28mm equivalent focal length, shot by an ordinary diner standing about 1-1.5 meters away. Auto exposure, auto focus, auto white balance. Retain slight hand shakiness, breathing fluctuations, half-beat delays in re-framing, brief focus searching, motion blur, local overexposure by the window, and shadow noise. No filters, no beauty mode, no skin smoothing, no cinematic lighting, no stabilizer movement, no artificial shallow depth of field. Retain real skin texture, stray hairs, clothing wrinkles, and natural reflection of thin socks. [Character Locations] Landlady #1: Left rear beige-grey double sofa. Leaning back browsing phone, natural Siren Pose, ignoring diners for now. Man #2: Front right first green old table, eating alone, the one who calls for a drink. Diner #3: Back right second independent table, about one meter from Man #2 with an aisle in between. Eating alone, not synchronized with Man #2, not speaking yet. [Space & Props] Left rear: Sofa, phone, a pair of black high heels by the feet, one closer to the camera. Center rear: Drink cooler, service counter, beige ledger. Front right table 1: Man #2, bowl of noodles, chopsticks, small seasoning dish. Back right table 2: Diner #3, another bowl of noodles, chopsticks, seasoning dish. Tables/props are independent and not shared. The cooler contains only one orange drink: ~500ml transparent plastic bottle, orange-yellow translucent liquid, colored tamper-evident cap. Only one bottle appears in the entire film, not on the table beforehand. [15s Shot List] 0-2s: Low-angle foot establishing shot starting from #1's legs on the sofa edge. 2-4s: Mid-shot of Landlady on sofa as camera moves up to her upper body. 4-6.8s: Wide mid-shot showing both dining tables with the aisle between them. 6.8-8.5s: Man #2 finishes a mouthful and shouts for a drink. 8.5-10.5s: Landlady responds, puts phone down on the armrest, puts on heels, and stands up. 10.5-12.5s: Mid-shot of Landlady taking one orange drink from the glass cooler and picking up a ledger, walking toward the aisle with a natural S-curve sway. 12.5-15s: Handheld tracking shot follows her to Man #2's table. She places the sealed drink: 'Here is your drink.' Final frame: Drink is sealed, landlady holds ledger, two men at separate tables. [Audio] Diegetic restaurant sounds: cooler compressor, exhaust fan, distant chatter, noodle slurping, chopsticks hitting bowl, heels on floor, cooler door, footsteps, bottle hitting table. Realistic reverb. No BGM. [Continuity] #1 always moves from left-rear to front-right. #2 and #3 remain at separate tables. Drink stays sealed until delivered. [Negative Prompt] Do not sit #2 and #3 together; no sharing bowls; no cloning characters. No dancing, winking, or erotic performance. No skin smoothing, no floating objects. Act 2: [Reference Lock] (Same character/clothing references as Act 1). [Starting Point] Continuation from Part A final frame: Landlady #1 stands in the aisle facing Man #2 with ledger; Man #2 touches the sealed drink; Diner #3 watches from his table. [Overall Setting] Man #2 asks for the price; hearing 6 yuan, he offers 5. Landlady calmly takes the drink back, opens it, takes a sip, and hands the opened bottle back. Diner #3 watches and says: 'I'll take a case of those.' Landlady breaks her cold expression and sprays the drink into the back of her ledger while laughing. [Realistic Filming] (Same technical specs as Act 1). [Shot List] 0-2s: Man #2 asks 'How much?' Landlady replies 'Six.' 2-3.8s: Man #2 offers 'I only have five, how about five?' 3.8-5s: Landlady silently takes the bottle back. 5-6.3s: She unscrews the cap with a 'click'. 6.3-7.8s: She takes a small sip, looking at #2. Liquid level drops. 7.8-9s: She loosely places the cap back and hands the opened bottle to the stunned Man #2. 9-11.8s: Diner #3 stops eating and says seriously: 'I'll take a case of those.' 11.8-15s: Landlady's eyes widen, she turns toward #3, holds up the ledger to cover her face, and sprays a fine orange mist of the drink against the ledger while laughing. Final frame: #2 stunned with the opened drink, #3 waiting for a case, Landlady laughing behind the ledger. [Audio] Cap clicking, liquid sloshing, swallowing, and Landlady's short choke/laugh at the end. [Negative Prompt] No merging tables; no floating caps; drink must not refill itself. No eroticized movements. No large projectile vomiting, just a laugh-induced spray. No platform UI or watermarks.

Style: Bright, energetic, vibrant TV/social media ad – fast-paced, upbeat, youthful, suitable for Nigerian audiences Duration: 10 seconds Visual sequence: Opening: Close-up of a glossy Fizzo bottle/can, condensation glistening, bright sunlight hitting it Action: Hand twists open the cap – crisp fizz sound, bubbles rising vividly Shot: Ice cubes drop into a glass, Fizzo pours in with a sparkling stream Scene: Group of smiling young friends laughing, sharing cold Fizzo at a hangout/park Closing: Product shot with bold text: FIZZO – CRISP FIZZ, EVERY SIP! Colors: Bright brand tones (red, blue, white), fresh, lively lighting Audio: Upbeat catchy jingle + crisp fizz sound effect

FORMAT 15 seconds | 16:9 | live-action military documentary footage Fictional aircraft-carrier emergency at sea. Hyper-realistic raw footage. Fast multi-shot storytelling across handheld, telephoto and deck cameras. STORY A carrier-based fighter returns over rough open water with an apparent mechanical problem. The flight deck enters emergency recovery mode. Crew clear the landing area as the fighter approaches and lands safely. 00–03s — EMERGENCY Wide handheld footage on an active carrier deck under heavy overcast skies. Strong sea wind, grey ocean, wet worn deck. Crew receive an alert and urgently clear the landing area. One crew member signals others away as the camera quickly turns toward the ocean. 03–06s — FINAL APPROACH Hard cut to long telephoto footage. A fictional fighter approaches low over the ocean, landing gear down. Aircraft makes subtle roll and yaw corrections against turbulent air. Heat distortion behind engines. Camera operator struggles slightly to maintain framing. 06–09s — DECK SCRAMBLE Shoulder-mounted footage among crew. Personnel rapidly move equipment and take safe positions. One crew member crosses foreground, briefly blocking the lens. Camera pans toward the approaching fighter. Controlled professional urgency, no theatrical panic. 09–12s — TOUCHDOWN Low deck camera from a safe side position. Fighter touches down firmly. Landing gear compresses, tyres produce a small realistic puff, aircraft rapidly decelerates. Camera makes a fast imperfect pan following it. No explosion or exaggerated sparks. 12–15s — SAFE STOP Three-quarter handheld footage. Aircraft reaches a controlled stop, engine still running and heat haze visible. Crew cautiously approach, signal toward the cockpit and inspect the aircraft. Final frame holds briefly on the fighter surrounded by responding crew with rough ocean behind. CAMERA Real military documentation/news footage. Mix handheld deck camera, shoulder-mounted footage, long telephoto tracking and fixed low deck camera. Hard cuts. Reactive pans. Foreground obstruction. Autofocus corrections. Telephoto compression. Wind-induced micro-shake. Natural motion blur. Camera operators react to events rather than anticipating them. ENVIRONMENT Working aircraft carrier in rough open ocean. Wet non-slip deck, painted markings, deck equipment, parked fictional aircraft, heavy clouds and strong maritime wind. PHYSICS Aircraft has convincing mass, momentum and aerodynamic behaviour. Approach remains aligned with the carrier. Landing gear compresses naturally. Deceleration has believable inertia. Crew clothing and straps react continuously to wind. Jet engines produce realistic heat distortion. AUDIO Powerful wind, ocean ambience, carrier machinery, shouted instructions, approaching jet roar, tyre contact, aircraft deceleration and crew communication. Sound perspective changes naturally between shots.

A photorealistic cinematic scene of a muscular grey bulldog chained to a tree in a lush green backyard on a sunny day A sleek grey cat hits the bulldog with a red metal tray shattering its teeth into white shards The cat sweeps the shards onto the tray and mocks the furious dog The bulldog lunges violently but a thick rope violently snaps him back to the ground A tiny brown mouse appears and points at the dog laughing The cat dances just out of reach The bulldog leaps forward and grabs the cat squeezing tightly but the cat magically slips out of a fuzzy grey sweater leaving the empty sweater in the dogs paws The cat is now wearing a tight red onesie The bulldog freezes completely still mid lunge like a statue holding the empty sweater The camera zooms into a black circular iris transition opening up into a lavish dimly lit royal dining room with velvet drapes The grey cat is now lounging on an opulent red throne ringing a small golden handbell with a smug expression The tiny brown mouse struggles to walk while balancing a gigantic silver serving platter loaded with massive colorful domed food dishes on its head Cinematic lighting volumetric dust particles 8k resolution highly detailed fur textures shot on 35mm lens lively orchestral comedy music playing throughout Seedance 20 15 seconds 16 9 aspect ratio

[Overall Style] Movie-grade realistic texture, pure ancient Chinese Xianxia aesthetics. The first half establishes tension through supernatural suspense, while the second half flips into a restrained deadpan comedy. Uses classical white-space composition, cold moonlight, warm lantern light, dark blue and faded ivory tones, fine film grain, and directional sound misdirection. No modern elements. [Characters] Character ID A | Senior Sister: East Asian female from @image 1, 25-30 years old, oval face, fair skin, dark almond eyes, black hair partially held by a jade hairpin, tall and slender, wearing white embroidered silk Hanfu. She enters the courtyard holding a silver sword and a bronze lantern. Character ID B | Junior Sister: East Asian female from @image 2, 20-25 years old, round face, braided black hair, small stature, wearing green linen Hanfu and a white thin gauze layer, holding a bamboo script. [Shot 1 | 0-5s | Low-angle wide tracking] A desolate midnight stage at a mountain gate with wet stone, paper screens, and bamboo. The sister from @image 1 enters as a woman's crying and wood-dragging sounds are heard. [Shot 2 | 5-10s | Medium tracking] The senior sister approaches a paper screen as a hunched female shadow rises behind it. The junior sister screams 'Give me back my life!' while the senior sister draws her sword in a defensive pose as the camera zooms into the shadow. [Shot 3 | 10-15s | Close-up reveal] The screen slides open to reveal the junior sister in her green Hanfu and white gauze, holding a script and a wooden gavel. She asks with a smile: 'Sister, do I look like a ghost?' The senior sister, still holding her sword, replies expressionlessly: 'Yes. Half a step later, you would have been one.' Junior sister's smile vanishes. The frame freezes on the senior sister rolling her eyes as the horror music stops and the gavel drops. [Technical Requirements] 15 seconds, 16:9 aspect ratio, three coherent shots, synced Mandarin dialogue, realistic movement of lanterns, shadows, and fabric. [Negative Prompts] Same as previous.

Grok, please turn the sand worm’s head because that is the way he burrows. And the way the sand-worm is guided by the people of dune climbing up on his back with hook ropes and then the sand worm keeps his head up to travel fast to keep the sand-worm...

Use @Image 1 as a strict reference for the identity, faces, hairstyles, and outfits of two individuals, keeping the black-haired idol on the left and the short brown-haired idol on the right. @Image 2 serves as environmental design material for a massive cyber live venue; do not include 2x2 grids or white borders. High-definition theatrical anime style with a transparent cel-look for characters and deep cinematic CG for the venue. Featuring blue, cyan, and purple lasers, audio waveform LEDs, a packed audience, and countless blue glow sticks. 0-1.8s: Start immediately at the peak of the chorus. Ultra-wide front view, the two are already on center stage singing and dancing in sync. Lasers and glow sticks explode in light on the first beat as the camera zooms in fast. [cut] 1.8-3.8s: Low-angle full-body shot, powerful symmetrical steps and arm swings synced to the beat. Hair, translucent sleeves, and ribbons sway with inertia. Cyan light waves spread from the floor with each step. [cut] 3.8-5.9s: Frontal close-up, camera glides between them as they take turns singing to the lens and wink together at the end. Lips move in perfect sync with the female duo vocals. [cut] 5.9-8.0s: Panning mid-shot, the two run to the front of the stage, reaching out to the audience. Glow sticks rise in unison and background LEDs react to the sound pressure. [cut] 8.0-10.2s: Center full-body shot, the two spin back-to-back. Camera rotates in a semi-circle as cyan and purple light lines on outfits create circular trails. [cut] 10.2-12.6s: Extremely low angle, the two jump simultaneously; a giant bluish-purple ring of light diffuses from the landing. Camera tilts up quickly to capture their smiles. [cut] 12.6-15.0s: Camera zooms out smoothly to show the arena panorama. The two pose together with arms wide. LEDs, lasers, and sparks reach peak brightness at the final beat. The original J-POP music ends with a powerful chord, leaving a short echo of cheers. Music: High-energy cyber J-POP at 160 BPM, 4-beat rhythm, synth arpeggios, bright female duo vocals, perfectly synced to the lighting and action. Restrictions: Maintain consistency for the two characters in every cut. No extra characters, no face/outfit changes, no limb deformities, no watermarks, or text.

[Overall Style] Deadpan supernatural reversal comedy, opening with real cinematic texture, disguised as an oppressive urban ghost story in the first half, revealing it as a common rainy night misunderstanding in the second half. Featuring cold cyan streetlights, warm light from a convenience store, reflections on wet asphalt, restrained facial performances, precise spatial staging, and a rhythmic transition from horror sound effects to comedic audio. [Characters] Character A | Sword Immortal Sister Use @Image 1 as strict identity and costume reference: same 25-30 year old East Asian female, oval face, sharp dark eyes, long straight black hair, fair natural skin, tall slender figure, wearing white embroidered silk Hanfu, translucent layered wide sleeves, silver waist ornaments, jade hairpin, and white cloth boots. Character B | Bike Sister Use @Image 2 as strict reference: same 25-30 year old East Asian female, maintaining the same face, ponytail, body proportions, clothing, accessories, and the same bicycle covered by a translucent white raincoat. [Shot 1 | 0-5s | Low Angle Wide Slow Tracking] 16:9, empty modern urban underground passage at night, heavy rain, concrete pillars, dripping pipes, flickering fluorescent lights, distant neon signs, scattered leaves, and heavy mist clearly visible. Character A stands fully in the light, holding the same silver sword. Character B silently approaches on the bike covered in the raincoat; the reflectors look like glowing eyes, a plastic bag flapping like a ghost's tongue. Low horror ambient sound, rain, and bike chain noise. [Shot 2 | 5-10s | Medium Shot] Character A suddenly turns, draws the sword, sleeves billowing, shouting: "Show yourself, monster!" The raincoat figure doesn't answer, gliding forward steadily. Camera tracks back at the same speed, horror frequency intensifies, then a small crisp bike bell rings. [Shot 3 | 10-15s | Close-up to Two-Shot] Character B brakes under convenience store light, lifts the raincoat hood, revealing the consistent face, saying expressionlessly: "Sister, I just got off work. You're the one who looks like a ghost." Heroic sword sound stops instantly. Extreme close-up: Character A looks at her white Hanfu, hides the sword behind her back, nods while trying to remain cool, one eyelid twitching slightly. Freeze on Character B staring at her. [Technical Requirements] Strict 15s duration, three shots, stable character identity, coherent motion direction, realistic physics for rain, hair, and fabric, native Mandarin dialogue and spatial audio, no subtitles. Optimized for Seedance 2.0 multimodal input. [Negative Prompts] blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props

Add two new elegant women dancing joyfully with the two dancers already in the first image, in the same vibrant Cuban plaza celebration for a free Cuba.

A stylish young woman walks through a crowded city street during golden hour. With a simple snap of her fingers, time freezes instantly—cars stop mid-motion, birds remain suspended in the sky, raindrops hang in the air, and every person becomes perfectly still. She calmly explores the silent world, walking between frozen moments with cinematic tracking shots and breathtaking visuals. She then notices a little girl moments away from being struck by a falling construction sign. While time stands still, she gently moves the child to safety. With one final snap, time resumes. The sign crashes harmlessly to the ground, the crowd continues as if nothing happened, and the mysterious woman quietly walks away with a smile. Ultra-realistic visuals, emotional storytelling, golden-hour lighting, seamless transitions, and a powerful cinematic ending.

[Overall Style] Cinematic realistic restrained reversal comedy, using elegant Xianxia cinematic language and precise spatial staging. Humor comes from status reversal. Designed for Seedance 2.0's character locking, multi-prop coordination, and native audio-visual synchronization. [Characters] Character A: 25-30 year old East Asian Sword Immortal Sister from @Image 1. Strict adherence to face, hair, and white silk Hanfu. She floats half a meter above ground with a sword rotating around her. Character B: 25-30 year old East Asian Bike Sister from @Image 2. Strict adherence to identity and bicycle. [Shot 1 | 0-5s | Low Angle Wide Slow Tracking] Modern Chinese city rooftop bike platform at sunset. Character A is floating proudly. Character B enters pushing her bike, looking in awe. [Shot 2 | 5-10s | Medium Shot] Camera orbits slowly. Character B asks: "Sister, can you really ride a sword?" Character A replies calmly: "A little." Character B rings her bike bell once. In the background, six bikes vibrate and ring back rhythmically. [Shot 3 | 10-15s | Wide to Close-up] The six bikes rise into a giant circular "Bicycle Sword Array" behind Character B. She steps onto her floating bike and says: "I usually commute with this," then glides away. Character A's sword hides behind her legs. Extreme close-up of Character A's frozen smile and twitching eyelid. [Technical Requirements] Strict 15s, three clean shots, stable faces/clothes, realistic physics for bikes/fabric, accurate Mandarin lip-sync, precise bell synchronization, no subtitles. [Negative Prompts] blurry, bad quality, low quality, low resolution, noisy, jpeg artifacts, watermark, text, error; deformed, mutated, bad anatomy, poorly drawn hands, bad composition, out of frame, disfigured; inconsistent character, changing clothes, face morphing, background shift, glitching cuts, disappearing props

Night retro old-town street, cinematic CG, high-fashion girl group MV mood, shallow depth of field, cool-warm color grade. 0-2s: Medium shot, slow arc left. [image2] stands under a streetlamp, lifts her eyes to camera. Focus locked on [image2]. 2-5s: Camera keeps arcing left, focus pulls forward. Blurred figure [image2] in the foreground resolves into a low-angle close-up, turning to face camera. [image2] blurs into bokeh. 5-9s: Camera pulls back, panning left. [image3] recedes to mid-ground. [image4] walks in from the left, filling the foreground. Focus shifts from [image3] to [image4]. 9-12s: Camera arcs around. [image4]falls back. [image1] walks in from the right into a foreground close-up. Focus moves to [image1]. 12-15s: Wide shot, camera pulls back and rises, then settles. Four members line up in a power stance — [image2] center, [image3] and [image4] on the sides, [image1] closing. Stone arches for depth, streetlamps forming a light trail, wet reflective ground.