Cinematic Realistic

Generate a 12-second, 9:16 vertical, realistic movie-level tech product advertisement based on the reference product image. The product must strictly match the reference image from the first frame to the last: a white desk fan, circular fan guard, ChatGPT six-petal vortex-shaped special fan blades, white rounded base, and the base must permanently maintain and only have 4 mechanical buttons. The four buttons are permanently fixed from left to right: 1st (leftmost) = go; 2nd (left-center) = plus; 3rd (right-center) = pro; 4th (rightmost and largest) = reset. This is a fixed product structure. The video is strictly prohibited from changing the number, position, size, arrangement, and text of the buttons. The highest priority is not recognizing the button text, but strictly executing the following physical space actions: leftmost button → move right one slot → move right one more slot → rightmost large button. Specifically: press the leftmost button first; press the 2nd button from the left second; press the 3rd button from the left third; finally press the rightmost largest button. The text go / plus / pro / reset is only for auxiliary recognition. [Key Interaction Rules] The first three button presses must be completed in the same continuous shot. From the first press of 'go' to the completion of the third press of 'pro': no camera cuts, no change in camera position, no left-right mirroring, no change in focal length, no orbiting, no regenerating the hand, no allowing the hand to leave and re-enter the frame. The camera is always directly in front of the product, looking slightly down at about 20°. Fully display the four buttons so they all exist in the frame simultaneously. A right hand enters from the bottom left of the screen. The index finger first hovers directly over the leftmost button. Afterward, the index finger can only move along a horizontal path from left to right. Returning to the left is strictly prohibited. The movement must be presented as a clear: PRESS → SLIDE RIGHT → PRESS → SLIDE RIGHT → PRESS. Move only the distance of one button each time. [0.0-1.2s | Product Opening] Pure white studio, white desktop. Fan is static. Front 3/4 product hero shot, showing the full fan. ChatGPT six-petal vortex blades clearly visible. Four buttons on the base clearly arranged: go | plus | pro | reset. Camera slowly dollies push-in. High-end tech product ad quality, soft studio lighting, natural shadows, realistic white plastic and metal materials. No human hand allowed at this time. [1.2-4.8s | Continuous Three-Stage Operation Shot] Hard cut to a fixed shot slightly looking down from the front of the product. No editing allowed from now until 4.8s. Simultaneously see: entire fan body, full base, full four buttons. Buttons fixed left-to-right: go | plus | pro | reset. A right hand enters from the bottom left. Important: hand must enter from the left, not the right. Index finger first stops directly over the 1st button 'go' on the far left. (1.4-2.1s): Index finger vertically downward, presses only the 1st button. Real mechanical sinking feedback occurs. Others remain still. Fan starts rotating at low speed. Index finger lifts again. (2.1-3.2s): Index finger stays in frame, slides horizontally right from the 1st button, moves one button distance. Stops over 2nd button. Vertically presses 2nd button. Fan speed increases. Index finger lifts. (3.2-4.5s): Index finger slides right again one button distance, stops over 3rd button, vertically presses it. Fan enters highest speed. (4.5-4.8s): Index finger lifts, hand leaves from bottom right. Fan keeps rotating at top speed. The first three operations must form a clear 'leftmost → second → third' visual trajectory moving left to right. [4.8-6.0s | Wind Performance] Switch to 3/4 side ad shot. Camera slides horizontally. Fan rotates at high speed. Vortex structure forms elegant motion blur. A thin white paper at the desk edge is gently moved by the wind. Restrained, high-end look. No hand in this segment. [6.0-7.2s | Blade Macro] Macro shot at very close range. Camera near fan center. Rotating six-petal blades form beautiful ChatGPT vortex. Central axis remains stable. Motion blur only on rotating structure. Real mechanical movement. No floating logos or 2D animation. [7.2-8.6s | Hero Shot] Low-angle 3/4 side. Fan runs at high speed. Camera slowly dollies out with slight orbit. White paper and silk ribbon sway. Button area remains: go | plus | pro | reset. No hand operation. [8.6-10.0s | RESET] Hard cut back to frontal fixed shot identical to 1.2-4.8s. Four buttons visible: go | plus | pro | reset. Hand enters from bottom right. Index finger appears directly over the large 'reset' button on the far right. (8.6-9.2s): Hover over large button for half a second. (9.2-9.7s): Index finger vertically downward, presses only the largest 4th button. Clear mechanical sinking. Others stay still. [10.0-11.0s | Stop] Reset is pressed, fan stops immediately.

Cup noodles correspond to Reference Image 1. The cat corresponds to Reference Image 2. The host corresponds to Reference Image 3. 0–2.5 seconds: Close-up shot. holds the unopened in one hand, brings it close to the camera, and shakes it excitedly while pointing at the cup with the other hand and talking directly to the camera. then tears open one corner of the paper lid, lifts it, and looks down into the cup. The camera quickly pushes in toward the opening. Inside, there is only a dry noodle block. nods with satisfaction. End state: the paper lid is half-open, and the are placed in the center of the table. 2.5–5 seconds: An electric kettle enters the frame from the right and pours hot water into the cup. The water stream is clearly audible and white steam rises from the cup. keeps talking while pouring. After finishing, presses the paper lid back down flat and pats it twice with the palm of the hand. End state: the paper lid is closed, with a thin stream of steam escaping from the rim. 5–7.5 seconds: takes out a white mechanical dial kitchen timer, forcefully twists the dial all the way to the end, then turns it back to the desired time and places it beside the cup noodles with a sharp tap. The scene immediately enters fast-forward mode: rapidly jitters and shifts around in the seat, talks, drinks water, checks the phone, and fidgets with exaggerated time-lapse-like movements. The lighting outside the window remains unchanged. The timer dial visibly spins rapidly back toward zero while steam continues escaping through the gap in the cup lid. End state: the dial reaches zero and the timer rings loudly while vibrating intensely. 7.5–10 seconds: The instant the timer rings, fast-forward mode stops and everything returns to normal speed. excitedly reaches forward and tears open the paper lid in one quick motion. At the exact same moment the lid flips open, the head of suddenly pops straight out of the cup. stares directly into the camera with huge wide-open eyes and a slightly open mouth; its ears flatten backward for an instant, then spring upright. At that exact moment, all sound suddenly cuts out. and stare at each other in complete silence for half a second. The camera rapidly pushes into an extreme close-up of the cat’s face. End state: ’s head is sticking out of the cup, while ’s hand is frozen in midair. 10–12.5 seconds: first looks left, then right, then looks up at with a completely confused expression. It hooks both front paws over the rim of the cup and pulls its upper body up. Its hind legs push against the inside wall of the cup, launching the cat out. lands with all four paws on the tabletop, then rapidly sprints toward the left side of the frame and completely exits the shot. While running, it knocks over the kitchen timer. The wobble from the impact but do not fall over. End state: only the empty cup and the fallen timer remain on the table. is completely

Cinematic dynamic action scene full length shot of a fierce female Korean martial artist with long, dark blue hair. She is wearing a modern black sports top, white Taekwondo pants, a black belt, and is barefoot She performs powerful high-energy Taekwondo roundhouse and spin kicks against multiple male opponents inside a dimly lit gritty industrial warehouse Cinematic volumetric lighting with dust motes floating in the air, high impact fight choreography, dramatic energy shockwaves and dust explosion effects upon impact, camera panning dynamically, 8k resolution photorealistic hyper detailed action movie sequence.

Duration: 10 seconds Aspect ratio: 9:16 Reference: @Image1, lock the protagonist's face, hairstyle, and styling, no additional description or changing of clothes Audio: Realistic city night ambient sound, running footsteps, strong wind whistling, low-frequency rebound of elastic device, muffled landing sound, friends cheering; no dialogue, no narration, no subtitles Overall style: Extreme sports mobile phone live recording, real high-altitude distance, strong sense of presence, one-shot throughout, slight handheld and high-speed motion blur, reject game CG feel GLOBAL CONTINUITY Setting: Nighttime city skyscraper rooftop, green helicopter landing pad, yellow circular markings, low concrete parapet. A vertical street canyon formed between two tall buildings, with vehicles, streetlights, and intersections visible on the ground. Lower apparatus: A huge circular black elastic rebound device is set in the center of the street, with a deep black elastic surface in the center, surrounded by a continuous rainbow light belt in the outer ring. The device is huge and directly opposite the edge of the roof, serving as the accurate landing point for the protagonist to fall and bounce. Main subject: @Image1 is the sole protagonist. Identity, face, and styling remain consistent throughout. Bystanders: Three to four friends standing on the left side of the roof, filming with mobile phones and waiting for the protagonist to return. They must not steal the focus from the main subject. CAMERA CONTINUITY The entire video is a one-shot filmed by a friend on the roof with a mobile phone. The filmer first follows the run-up behind the protagonist; after the protagonist jumps, the filmer quickly rushes to the parapet to look down; then the camera continues to lock on the protagonist, gradually lifting from vertical downward to the sky; after the protagonist bounces back to the roof, the camera returns to eye level. No cuts, no teleporting, no changing perspectives. STUNT ENGINE The action loop must be fixed: Rooftop run-up → Leap over parapet → High-altitude fall → Hit the center of the circular elastic surface → Deep depression of the device → Strong rebound → Protagonist rises at high speed → Mid-air tuck and flip → Positioning for landing → Feet return to the same rooftop → Knee-bend landing → Cheering celebration. SHOT 1 (00:00-00:10) High-altitude Rebound Stunt Subject: @Image1 protagonist stands at the back of the rooftop helipad, facing the distant parapet. Friends are on the left side of the frame filming. Action: 00:00-00:01.50 | Fast Run-up The protagonist's back is to the camera, running quickly from the center of the helipad to the edge of the roof. Strides gradually increase, body leans forward, arms swing naturally. The filmer follows closely behind the protagonist with a mobile phone, creating realistic footstep vibrations. A friend stands on the left holding up a phone, watching tensely. 00:01.50-00:02.30 | Leap Off Rooftop The protagonist takes a powerful final step, leaping over the low parapet. At the moment of leaving the roof, arms spread to the sides, body leans slightly forward, actively diving into the gap between the two tall buildings. The filmer rushes to the parapet, the camera quickly tilts downward, always keeping the protagonist in the center of the frame. 00:02.30-00:04.80 | High-altitude Fall The camera is almost vertically downward. The protagonist descends rapidly through the space between the buildings, body adjusting from a forward lean to facing downward, arms staying spread to stabilize direction. As the protagonist descends, streets, vehicles, and intersections quickly appear. A giant black circular elastic device appears from the bottom of the frame, the rainbow light ring illuminating the surrounding buildings. The protagonist's descent trajectory must accurately target the center of the black elastic surface, not hitting the light ring or the buildings. 00:04.80-00:05.50 | Hitting the Center The protagonist's body lands right in the center of the black elastic surface. The elastic surface clearly forms a deep depression after the impact, the rainbow light ring remains fixed, and the device must not break. After a brief compression, the elastic surface returns to its original state with huge force, popping the protagonist vertically upward. Must clearly show the four stages: "contact—depression—compression—rebound," no premature reverse flight before contact. 00:05.50-00:08.20 | High-speed Rebound and Mid-air Flip The filmer is still standing at the rooftop parapet. The filmer's feet briefly appear at the bottom of the frame, reinforcing the realistic first-person perspective and high-altitude distance. The protagonist rises at high speed from the center of the circular device, returning along the original vertical path to the roof. The camera continues to lift from vertically downward to upward, following the protagonist through the space between the city buildings. After taking off, the protagonist tightens the abdomen, knees close to the chest, arms holding the shins, forming a tight tuck, and completes a continuous rapid flip in the air. Do not lock a specific rotation direction, prioritize natural continuous rotation. As the protagonist approaches the roof height, they begin to open the body: release the legs, look up to find the roof, turn feet downward, and spread arms for balance. 00:08.20-00:09.40 | Return to Rooftop The protagonist approaches the camera at high speed, leaping back onto the helipad from outside the parapet. Both feet contact the green ground simultaneously, knees and hips bend deeply to absorb the landing impact in a low squat. The landing must have clear weight: soles or feet contact, legs compressing, body sinking, slight rebound to recover. Must not land standing straight up or lightly. The filmer quickly takes half a step back to avoid the protagonist hitting the camera while maintaining a frontal close-up. 00:09.40-00:10.00 | Realistic Celebration The protagonist quickly stands up from the squat, looking excitedly at the camera and cheering. Rooftop friends rush into the frame, surrounding the protagonist with phones, patting shoulders, punching the air, and celebrating loudly. The camera shakes slightly due to the crowd approaching, ending in a realistic mobile phone selfie video state. Environment: Tall building lights, street vehicles, and the rainbow light ring maintain spatial stability. The fall and rise must happen along the same vertical line, and the protagonist eventually returns to the same rooftop they jumped from. Camera: 24mm mobile phone wide-angle, one-shot. Begins with a low-position follow-up from behind; after jumping, becomes an over-the-shoulder downward shot; during rebound, continuously lifts from vertical downward to upward; finally returns to eye-level close-up. Style: Realistic night mobile phone exposure, tall buildings have accurate perspective and scale. Wind noise increases during the descent phase, and city lights form natural motion trails; the protagonist returns to the roof extremely fast, but the human silhouette remains clear. Constraints: This is a professionally arranged fictional extreme sports scene. Must fully show the circular elastic device catching and rebounding the protagonist, not showing a free fall without protection. The protagonist's descent and ascent must be continuously traceable. The rebound comes from the physical compression of the black elastic surface, not flight, teleportation, rope pulling, or superpowers. Negative: Changing @Image1 character styling, redesigning clothes, clothing drift, face drift, character replacement, mid-cut, camera teleportation, scene jump, filmer suddenly appearing on the ground, third-person aerial photography, protagonist flying directly, floating, invisible rope lift, anti-gravity, instantly returning to the roof, no elastic device, wrong landing point, landing on rainbow light ring, hitting buildings, hitting parapet, rebounding before contact, elastic surface not sinking, device breaking, device deformation and disappearance, descent direction change, ascent path offset, body passing through elastic surface, abnormal character size, distorted limbs during flipping, extra arms, extra legs, joint inversion, body melting, duplicate characters, landing without weight, landing standing straight, feet penetrating the ground, falling and injured, blood, serious injury, friends blocking the camera early, crowd frozen, phone deformed, tall buildings bent, floating vehicles, anime rendering, game CG, plastic skin, cheap green screen, excessive lighting effects, excessive slow motion, subtitles, screen text, watermark, brand logo, voiceover, extra dialogue

Ultra-realistic cinematic scene of a humanoid robot kneeling beside a young child on the floor, carefully helping with a small toy project. Dark cozy room, dramatic warm flashlight beam, cool blue window light, realistic white-and-black robotic body, emotional storytelling, detailed reflections, shallow depth of field, cinematic lighting, photorealistic, 8K, high detail, film still aesthetic, natural composition, vertical.

A cinematic 15-second monochrome lifestyle video featuring a young Asian woman [use reference image] in an oversized white shirt and oversized trousers, moving naturally through a modern house and garden. • Scene 1 (0–3s): She wakes up and sits on the edge of the bed, sunlight softly falling across her face as she looks toward the window. • Scene 2 (3–6s): In front of a mirror, she casually fixes her messy hair, smiling subtly at her reflection. • Scene 3 (6–9s): She walks barefoot through the kitchen, playfully spinning a mug in her hand before taking a sip. • Scene 4 (9–12s): Outside, she walks through the garden while gently touching the leaves and flowers along the path. • Scene 5 (12–15s): She sits beside the swimming pool with her feet in the water, looking peacefully toward the camera as the scene fades out. Cinematic black-and-white photography, soft natural lighting, subtle film grain, realistic skin texture, candid expressions, intimate framing, gentle handheld camera movement, smooth transitions, editorial fashion-film aesthetic, minimal and elegant atmosphere.

he tips forward and slides into it headfirst in one smooth impossible motion, the hat rocking once and settling upright as if nothing happened

exhales a long slow drift of glowing golden mist over the open ocean, a soft ribbon of dawn-colored vapor

a young man dressing up, cut to him wearing his jacket, cut to him walking out the door, cut to him walking in the streets, cut to him opening a sofa shop.

Ultra-realistic infinite zoom cinematic sequence, mind-bending visual transformation, premium VFX realism. One seamless continuous shot with a constant forward zoom, smooth acceleration, perfect match-cuts and impossible scale transitions. Sharp detail, cinematic depth, subtle motion blur, premium lighting evolution, surreal but photorealistic, no cartoon look. One adult human figure appears only at the beginning as the starting visual anchor. After that, the sequence becomes a visual journey through environments, structures and worlds. No dialogue. The sequence begins in a modern city apartment, then transforms through multiple scales and environments: human eye, room, city, clouds, Earth, space, galaxy, nebula, microscopic world, cell-like structures, abstract geometric tunnel, then a final new city reveal. Perfect seamless infinite zoom illusion, coherent scale transitions, photorealistic environments, macro-to-cosmic continuity, elegant match transformations, stable geometry, premium VFX detail, smooth camera flow, visually viral, no cuts, no text. 0-5s: Extreme macro push-in Start on an ultra close-up of a human eye. The camera zooms into the pupil. The reflections inside the eye reveal a window, then the entire apartment room appears inside the iris as the camera keeps pushing forward. 5-10s: Continuous infinite zoom The apartment becomes a city block seen from above, then the entire city, then clouds, then the curvature of Earth. The zoom continues smoothly into outer space, with the planet shrinking into a starfield. 10-15s: Acceleration through cosmic scale The camera passes through a spiral galaxy, then a glowing nebula, then dives into a bright star that transforms seamlessly into a microscopic cell-like structure. The cell membrane becomes a huge circular landscape. 15-20s: Final impossible reveal The camera continues forward through abstract crystalline patterns and geometric light tunnels, then emerges into a completely new futuristic city at sunrise, revealing that the entire cosmic journey has resolved into another world. End on a breathtaking wide reveal.

Create an ultra-realistic premium UGC perfume advertisement featuring a strikingly beautiful young adult woman in an elegant, bold, fashion-forward outfit. She has a naturally captivating appearance, expressive eyes, polished makeup, luminous realistic skin texture, and beautifully styled hair. Her look is sophisticated and confident rather than overly glamorous. Scene: A luxurious modern interior with warm ambient lighting, subtle shadows, and an upscale editorial atmosphere. Keep the background clean and slightly out of focus so the woman and perfume remain the visual focus. Outfit: She wears an elegant, bold evening outfit with a sophisticated silhouette—fashion-forward, refined, and alluring while remaining tasteful. The outfit should complement the luxury positioning of the perfume. UGC performance: She holds the perfume bottle naturally and speaks directly to the camera as if recommending a fragrance she genuinely loves. Start with a casual attention-grabbing moment, then show the bottle close to the camera, spray the perfume on her wrist/neck, smell it, and react with a subtle confident smile. Include natural hand gestures, slight movements, authentic facial expressions, and relaxed body language. Camera: Handheld smartphone-style UGC footage mixed with a few polished close-up shots. Natural micro-movements, realistic autofocus, shallow depth of field, occasional subtle push-in. Avoid the overly perfect look of a traditional commercial. Product focus: Give the perfume bottle several hero moments with crisp focus, realistic reflections, premium glass and liquid details, while keeping the brand label clearly visible and undistorted. Lighting & aesthetic: Warm luxury lighting, soft highlights on her face and hair, realistic skin pores and fine details, cinematic but believable exposure, sophisticated neutral tones, premium beauty-ad aesthetic. Video style: Authentic high-end UGC × luxury fragrance campaign, natural and aspirational, confident feminine energy, photorealistic, cinematic realism, 4K, realistic motion, natural expressions, no CGI appearance, no beauty-filter effect. Ending: She looks directly into the camera, holds the perfume beside her face, gives a subtle confident smile, and finishes with a natural recommendation-style expression. Avoid exaggerated acting, unnatural body movements, plastic skin, excessive retouching, distorted hands, warped perfume bottles, fake-looking reflections, or artificial facial expressions.

Aspect Ratio: 9:16, following the original composition Shot form: Single shot, one-shot, no transition Overall atmosphere: Quiet daily life after bathing, real and natural, restrained and relaxed, no intentional performance ## First Frame Lock The 0th second must accurately connect with `@Image 1`: maintain the adult female's face identity, skin color, body proportions, wet black hair, light grey-purple thin strap house dress, cream-white bath towel on the shoulder and existing standing posture. The right hand stays at the wet hair on the back of the head, and the left hand hangs naturally; the head is toward the left of the screen, eyes looking out of the bathroom. Retain small water droplets on the skin and collarbone, the weight of wet hair in bundles, and the cold-warm contrast formed by the bathroom cold light and bedroom warm light. ## 0.0–0.8 seconds | Breathing after still The screen starts accurately from the original image, without composition jumps. The person maintains the original posture, and the chest produces very slight fluctuations with a natural breath. The gaze still stays on the left side of the screen, the expression is calm, and the lips are naturally slightly parted. The right hand fingers are still in contact with the wet hair, only letting the fingertips produce slight adjustments. The wet hair and bath towel hang naturally under gravity and do not swing suddenly. The lens is fixed, only retaining a very slight real handheld breathing feel. ## 0.8–2.8 seconds | Tidying up wet hair The person's right hand fingers slowly pass through the wet hair on the back of the head, gathering a small bunch of hair back. The movement is gentle and continuous, without shaking the head forcefully. The head tilts slightly to the right with the hand movement, the amplitude is very small; the gaze still stays on the left, not looking at the camera in advance. The moved wet hair follows the finger movement with a half-beat delay, and a few strands of wet hair sticking to the side of the face and neck naturally separate. A small water droplet slowly slides down the side of the neck, reflecting briefly in the warm light. The bath towel only produces slight wrinkle changes and always stays stable on both shoulders. ## 2.8–4.6 seconds | Slowly withdrawing gaze After finishing gathering the hair, the right hand stops at the junction of the back neck and hair, not immediately putting it down. The person first slowly moves their gaze from a distance back to a position in front and slightly to the left, and then the head follows and turns back slightly. She naturally blinks once and exhales softly, the shoulders and neck relaxing slightly with the exhalation. The expression does not make an obvious smile, only retaining the quietness, relaxation and a bit of thoughtfulness after bathing. The lips naturally close after exhaling. The lens starts to zoom in very slowly, with a total advancement amplitude of no more than 5% of the original frame within six seconds, without changing the character's main body proportions. ## 4.6–6.0 seconds | Movement settled The right hand slowly slides from the wet hair to the back of the neck, finally stopping gently at the edge of the bath towel, not completely putting it down. The person maintains a relaxed standing posture, gaze falling near the left side of the camera, but not looking directly at the camera. The left hand fingers relax naturally, and the center of gravity of the body only undergoes slight adjustments. The wet hair naturally sticks back to the side of the neck after the movement ends, and the bath towel wrinkles slowly stabilize. Finally, stay for about half a second to form a quiet, life-like ending screen. ## Sound Design Retain the low-frequency operation of the bathroom exhaust fan, small dripping sounds in the distance, slight friction sounds of wet hair and fingers, and soft fabric sounds when the bath towel moves. No dialogue, no narration, no background music, no subtitles. ## Continuity and Prohibited Items - Character face, features, skin color, body type and sense of age are consistent throughout. - Wet hair length, color and wet state remain unchanged, not suddenly becoming dry, increasing or changing hairstyle. - House dress color, cut, strap position and coverage state remain stable, not sliding or deforming. - Bath towel always stays on both shoulders, not disappearing, changing color or suddenly falling off. - Water droplets can only slide down slowly, not suddenly increasing, flashing or all disappearing. - Maintain the door frame, bathroom, bedroom bed and cold-warm light direction in the original image. - Do not add new characters, mirror reflections, mobile phones, jewelry or other props. - Do not use head shaking, hair-flicking posing, exaggerated blinking, sudden smiles or direct looks at the camera and other intentional performances. - Do not zoom in on parts of the body, do not change the natural life-like shooting perspective of the original image. - Avoid facial drift, finger deformities, extra limbs, clothing clipping and background structure deformation.

[10-second manual pencil sketch drawing time-lapse] In professional digital drawing software, starting from a completely blank cream-white paper canvas, use a mouse cursor visible throughout the process to gradually reconstruct the provided reference image. Maintain the characters, poses, composition, long hair, and clothing of the reference image, as well as the scene relationships like the lake surface, arched bridge, distant mountains, willows, and pleasure boats. 0:00-0:02 Start with a blank canvas; the cursor quickly establishes character proportions and overall composition, outlining characters, lake surface, arched bridge, distant mountains, and main environmental contours. 0:02-0:04 Quickly refine the line art of characters and background, drawing faces, features, long hair, and clothing; simultaneously add bridges, mountains, willows, pleasure boats, and background figures. 0:04-0:06 Use a graphite pencil for quick cross-hatching to establish light and shadow on hair, face, clothing, and environment. Retain natural cross-hatching, repeated lines, broken lines, and manual correction marks. 0:06-0:08.5 Continuously refine facial features, hair strands, clothing folds, and environmental details to gradually form a complete manual pencil sketch. 0:08.5-0:09.5 Complete final details, erase a small amount of extra construction lines, and stably display the finished work. 0:09.5-0:10 The cursor moves to the bottom left corner of the screen to complete the final signature '● DeepBlue' with natural handwritten strokes. The signature can only appear at the end, and the cursor stays after completion. Style: High-speed manual pencil sketch time-lapse photography; cream-white paper, graphite lines, natural hatching, cross-hatching, paper texture, realistic hand-drawn marks. The canvas is the visual subject, with a simple and credible UI. Core Requirements: Blank canvas → Composition → Sketch → Detailed line art → Hatching light and shadow → Refinement → Completion → Final signature. All screen changes must be gradually produced by the visible cursor's actual drawing, and subsequent frames must be built on the previous stage. Prohibit characters or backgrounds from suddenly appearing, prohibit direct switching to the finished image, prohibit color painting, watercolor, 3D, anime style, and cinematic transitions.

10 seconds, cinematic, ultra-realistic, photorealistic, colossal scale, no text, no logos. The black sphere pulses. The frozen city suddenly moves again. On a rooftop far below, a young woman in a dark futuristic raincoat stands completely still, drenched by the rain. She looks upward as the sphere begins opening like a gigantic black eye. Her hair and coat whip violently in the gravitational wind. She slowly raises one hand toward the sphere. Inside it, the tiny city begins moving backward through time—cars reverse, rain rises, broken buildings reconstruct themselves. The woman watches in disbelief as her own reflection appears inside the sphere. The reflection moves independently. The real woman lowers her hand. The reflection doesn't. It suddenly turns toward her and points behind her. She turns around. Every skyscraper in the city is now facing her, windows illuminating one by one in a massive wave. The gravitational force suddenly intensifies. She is pulled across the rooftop toward the sphere, sliding through the rain. She grabs a massive steel structure, barely holding on as buildings around her stretch toward the black sphere. Extreme close-up of her eyes. Inside her pupil, the entire futuristic city is reflected. Then the reflection reveals something impossible: another version of the woman standing directly behind her. A deep pulse. HARD CUT TO BLACK. Sound: violent rain → deep gravitational rumble → metallic tearing → sudden silence → one sharp breath → enormous sub-bass pulse. Character: realistic young woman, dark futuristic raincoat, wet hair, natural facial features, cinematic lighting, determined but frightened expression. Keep her visually consistent throughout the sequence

Use Image 1 as the character reference for Alin, and Image 2 as the handgun reference. Create a 15-second vertical video, 9:16 canvas, with a horizontal cinematic frame centered inside it and black bars above and below. Scene: an eerie Eastern fantasy ruined town, burned wooden gate, dead branches, sparse peach blossoms, smoke, embers, fallen zombies on the ground. Alin: young East Asian woman, pale skin, delicate oval face, cold clear eyes, black hair in an ornate updo. She wears a pale jade-green hanfu with translucent wide sleeves, white and soft pink inner layers, gold peony embroidery, gold waist ornaments, pearl chains, and an ornate gold floral hair crown with jade, pearls, white flowers, and dangling tassels. She holds two black heavy semi-automatic handguns matching Image 2: square black metal slide, long barrel, textured black grip. Motion core: Alin does not perform fixed dance steps. Her movement is one continuous unstable improvised body flow, like a possessed dance. Tiny shoulder twitches travel through elbows, wrists, waist, hips, knees, and feet. She tilts her head, drops a shoulder, twists her waist, missteps, leans back, spins, sinks, and rebounds. Every motion flows from the previous momentum, no hard cuts, no frozen poses. She barely aims; her arms are carried by body inertia, and each casual wrist flick or turning motion fires a shot. Zombies fall back, collapse, or slide out of frame. No gore. Facial emotion progression: 0-3s blank and trance-like; 3-6s strange faint smile, unfocused eyes; 6-9s excited and dangerous; 9-12s suddenly cold and precise; 12-15s quiet, tired, with a disturbing slight smile toward camera. Timeline: 0-3s: Alin stands under the burned gate, head lowered, pistols hanging down. Subtle shoulder twitches begin, sleeves and tassels sway. Zombies approach from both sides; without looking up, she leans sideways and casually shoots one down. 3-6s: Her body becomes more erratic, hips shift off-balance, feet slide, torso collapses the opposite way. A strange smile appears. Her pistols draw loose arcs; short shots drop zombies on both sides. 6-9s: She half-spins, abruptly pauses, sleeves circle around her. Her head snaps with the rhythm. Zombies leap from behind or above; she flicks both pistols upward without looking, and they fall out of the air. 9-12s: She sinks low, rebounds upward, face turns cold. Arms cross, pistols sweep from low to high. Each passing muzzle triggered a short shot; zombies collapse one by one. 12-15s: The movement resolves. She slowly straightens, still faintly twitching. One final wrist turn drops the last zombie at the edge of frame. Pistols lower, sleeves settle, Alin looks into camera with a tired eerie smile. Music: dark Eastern ritual dance track, unstable low drums like broken heartbeat, guzheng plucks, pipa scrapes, low war drum, reversed breath-like vocals, tiny bell

Cinematic slice-of-life video, 30 seconds, single continuous narrative, photorealistic, natural realistic visuals, soft overcast daylight color grade, light rain throughout, smooth steady handheld-style camera, professional small clean subtitles burned in at the bottom of the frame, precisely synced to spoken dialogue, natural understated performances — no over-acting, calm normal conversation. 00–07s: A woman in her late 20s stands at a bus stop under a small umbrella, light rain falling, quiet street ambience. A man around the same age walks up, no umbrella, hood up, slightly damp. He stands a little apart, glancing at the rain. Dialogue (subtitled): Him: "Of course it rains right when I forget my umbrella かのんくん" Her: (small smile, shifting the umbrella slightly toward him) "Come here, there's room ." 07–14s: He steps under the umbrella beside her, natural slightly awkward-but-warm moment, both looking out at the rain. Him: "Thanks. Didn't expect that." Her: "I'd want someone to do the same for me." 14–21s: Easy natural conversation continues, small genuine laughs, comfortable silence in between. Him: "So do you take this bus every day?" Her: "Most days. You?" Him: "First time, actually. New job starts today." Her: "Good luck. You'll be fine." 21–27s: The bus approaches in the distance, headlights visible through the rain. Both glance toward it, a small shared look, like neither really wants the moment to end. Him: "This is us, huh." Her: (smiling) "Guess so." 27–30s: The bus doors open, they step forward together, she closes the umbrella, quiet warm nod between them as they board. Camera holds on the empty bus stop for a beat as the bus pulls away, rain continuing softly. Fade out. Camera: Smooth handheld-style realism, natural eye-level framing, subtle steady movement, gentle push-in during dialogue, documentary-like pacing, no dramatic angles モイネロ #SideM特別生配信 . Lighting: Soft overcast daylight, muted natural tones, light rain reflections on pavement, realistic wet-weather atmosphere. Subtitles: Small, clean, professional white sans-serif font, centered bottom third, subtle semi-transparent dark background box, precisely synced, simple fade in/out only. Audio: Natural rain sound throughout, distant traffic, umbrella rain patter, footsteps on wet pavement, bus engine approaching and doors opening, calm natural dialogue delivery, very light ambient acoustic music underneath at low volume, no dramatic score. Style: Realistic, warm, natural color grading, authentic candid feel, mature grounded tone, no exaggerated expressions.

Use the uploaded image as the only person reference, strictly maintain the person's features, face shape, hairstyle, clothing, body proportions, lakeside grass background, and overall lighting, no face swapping, no change of person identity. Duration 8 seconds, one-shot, realistic friend photo-taking behind-the-scenes feel. 0-2 seconds: The girl maintains the photo-taking pose in the original image, arms naturally crossed, body slightly sideways, maintaining a beautiful but restrained photo-taking smile on her face, eyes looking near the camera. There is a light breeze by the lake, long hair and the hem of the pink shirt naturally swing. 2-4 seconds: After maintaining the pose for a while, her eyes start to show a bit of inquiry, the smile slightly relaxes, she gently tilts her head to look at the photographer, and naturally asks: 'Is it done?' 4-5 seconds: After asking, she doesn't move immediately, still maintains the original photo-taking pose, quietly waiting for the other party's answer, eyes looking at the camera, with a bit of expectation. The off-camera photographer naturally replies: 'Done.' 5-8 seconds: After hearing 'Done', she clearly confirms the end of the shooting, and the whole person relaxes instantly. Her shoulders drop slightly, the originally crossed arms naturally loosen, the business-like smile on her face disappears, replaced by a truly relaxed smile. She exhales softly, the body sways slightly, her gaze moves to the side, and then she can't help but laugh, as if she finally doesn't have to continue posing. Key points: Must show the causal relationship of 'relaxing only after hearing done'. Before 'done', the person always maintains the photo-taking posture and does not relax in advance. The first half is an exquisite photo-taking state, and the second half is a very real life state, forming a clear but natural contrast. Requirements: The person's face is always stable, maintaining the same features as the uploaded image; 'Is it done?' Chinese lip-sync is natural and accurate; the off-camera voice 'Done' comes from the photographer, don't let the girl say it herself; no exaggerated performance; no large movements; hair moves naturally with the wind; retain real skin texture; the lens has a slight handheld feel; overall it's like a real behind-the-scenes clip captured by a friend while helping her take photos.

30 seconds | 16:9 | 4K UHD | 24fps Ultra-realistic AAA cinematic fantasy action Pure visual storytelling No dialogue Continuous one-take. REFERENCE LOCK: @image1 controls the hunter's identity, outfit, face, hair and furnace weapon. @image2 controls the frozen environment and ice creatures. Do not redesign either reference. VFX RULE: All creature destruction happens only after physical weapon contact. No spontaneous burning or unexplained explosions. No blood or biological gore. 0.0s–5.0s: Camera races low across cracked black ice toward the hunter. Thousands of ice beasts charge toward her. She grips the massive furnace weapon with both hands and performs one heavy downward strike. Contact creates a circular orange-white thermal shockwave that instantly cracks the ice armor of nearby creatures. 5.0s–10.0s: She slides across the frozen ground while dragging the weapon behind her, carving a glowing molten line through the ice. She pivots and swings upward. A fan-shaped wave of compressed heat travels through the creature formation. Impacted armor glows from within, fractures, turns black and collapses into steam, ash and glowing fragments. 10.0s–15.0s: A gigantic armored ice beast leaps from the background. Hunter blocks its claws with the furnace weapon. The collision creates a violent burst of steam and shattered ice. Camera circles around both figures while maintaining one continuous shot. 15.0s–20.0s: Hunter pushes the creature backward, then runs along a collapsing frozen wall. She jumps from the wall and lands directly on the creature's shoulder armor. The weapon reaches maximum temperature, mechanical pistons extending and molten channels turning white-hot. 20.0s–26.0s: She raises the weapon above her head and delivers one enormous vertical strike from shoulder to torso. The exact contact point becomes a glowing thermal mark. Cracks of white-orange heat spread across the creature's armor. The entire body transforms into black volcanic ash, white steam and burning armor fragments without showing flesh. 26.0s–30.0s: The frozen battlefield becomes covered in drifting steam and falling ash. Hunter lands on one knee, slowly rises, rests the giant weapon over her shoulder and walks toward a gigantic ancient frozen gate. Camera smoothly pulls backward, revealing the enormous ruined landscape. Negative Prompt: blood, gore, flesh, organs, wounds, red liquid, severed bodies, dismemberment, spontaneous fire, blue magic, ocean, city, green grass, desert, duplicate character, duplicate weapon, sword morphing, weightless weapon, katana, dual weapons, outfit change, hair change, face distortion, extra limbs, extra fingers, fused hands, weapon clipping, text, subtitles, watermark, logo, UI, hard cuts, freeze frames, slow motion, black screen, camera teleportation.