# Basic Specifications Duration 20 seconds, 16:9 horizontal screen, Chinese dialogue, ultra-realistic live-action film quality. Genre: Urban life slice × Light relationship suspense × Autumn atmosphere × Natural dance movements × Unfinished story. The entire video should not pursue exaggerated spectacle. The core goal is: **Real, natural, good-looking, strong sense of relationship, making the audience want to know what happened between these two people before.** Core event: The female lead is at home alone, casually tidying up and getting ready to go out. Someone she knows very well suddenly sends a message from off-screen: **"I'm downstairs."** She clearly didn't expect him to actually come. Then all actions revolve around: Surprise → Instinctive happiness → Deliberately pretending to be calm → Quickly getting herself ready → Going to see him. Finally, through an old object belonging to their past, a suspense is left behind. --- # Character Lock @Image 1 is only responsible for locking the adult East Asian female lead: Facial features; Face shape; Eyes; Nose; Lips; Skin tone; Hair color; Age impression; Overall body proportions. Throughout the entire film it is always the same adult woman. Female lead fixed: S1. Off-screen adult male fixed: S2. S2 never shows his face throughout. Male voice is natural, young, familiar, not a CEO-type deep voice. The two speak like people who have truly known each other for a long time. No broadcast announcer tone. No exaggerated short-drama tone. --- # Overall Visuals Early autumn evening. A real small apartment with traces of life but not messy. Not a mansion. Not a photography studio. Not a tech space. The space has: Warm gray walls; Natural wood wardrobe; A magazine left open casually at the foot of the bed; A canvas bag; A half-open window; Curtains gently blown by the wind; On the tabletop are a phone, earphones, a hair clip, and a lip glaze. Outside the window is natural warm light close to sunset. No complex artificial lighting effects indoors. Overall color palette: Cream white; Smoky brown; Olive green; Deep denim blue; A small amount of dark red. Natural, relaxed, with the warmth of life. --- # Brand New Costume System Completely avoid the previous: Preppy style; Black suit; Formal gown; Functional uniform; Racing suit; Dance training outfit; Eastern haute couture. ## Opening Home State Female lead wears: Washed old gray-blue loose short-sleeve T-shirt; Cream white soft loungewear pants; Barefoot or light-colored short socks. Clothing has slight natural wrinkles. Like she has really been at home for half a day. No feeling of "specially dressed up for the camera." --- ## Going-Out Look A natural costume-change cut is completed through wardrobe occlusion. The female lead then wears: Cream white fitted square-neck knit top; Olive brown cropped suede jacket; Deep indigo high-waisted straight-leg jeans; Dark red-brown vintage thin-soled shoes; Small dark brown shoulder bag. No high heels. No stockings. No short skirt. No workplace queen vibe. Overall it is: **2026 autumn urban relaxed outfit + a touch of vintage feel.** --- # Makeup and Styling Opening: Close to natural bare-face state. Real skin texture. Natural eyebrows. Lips only with a faint natural color. Black long hair casually half-clipped up with a large matte hair clip. Loose strands near the ears. --- Going-out state: The character's identity remains completely unchanged. Only very natural quick tidying: Remove the hair clip; Black hair naturally falls; Then uses her hand to tuck one side of her hair behind her ear; Applies a layer of low-saturation brick rose lip glaze; Wears a pair of very small old gold earrings. No sudden transformation into refined heavy makeup. The key point is: **She just suddenly became a little more serious.** --- # Camera Principles The entire film has 3 shots. The shots have a sense of life, but cannot shake like discarded casual footage. A small amount of real handheld breathing is allowed. No commercial-ad-style large orbiting. No meaningless showing off of technique. ## Shot 1 35mm feel. Natural medium shot in the bedroom. Very slight handheld. Function: Establish the state of life and the first hook. ## Shot 2 32mm feel. Naturally enter a short lateral tracking from wardrobe occlusion. Function: Character tidying herself + a very natural small segment of body rhythm. ## Shot 3 40mm feel. Corridor follow shot → elevator stops. Function: Relationship suspense and the ending old object. --- # SHOT 1 ## 00:00-00:05.5 ### "Didn't you say you weren't coming?" At the opening, the character should not look at the camera. Female lead S1 sits half on the edge of the bed with her back to the camera. Phone placed beside her. Evening wind outside the window moves the thin white curtains. A light, slightly retro Chinese indie pop song plays in the background. She lowers her head and slowly clips her hair back up. Completely an ordinary action in life. The phone suddenly rings once. Not an exaggerated notification sound. A voice message plays automatically. S2 off-screen voice: [Chinese] I'm downstairs. The female lead's hand movement immediately stops. Stops for about 0.4 seconds. This is the first real point of tension. She slowly turns her head to look at the phone. Her expression is not shock. First it is disbelief. Then a bit of happiness that is clearly hard to hide appears in her eyes. But she quickly suppresses it. She picks up the phone. S1: [Chinese] ...Didn't you say you weren't coming? Tone: Surprised; A little reproachful; But more than anything, actually happy. No coquettish tone. S2 does not answer immediately. The female lead looks at the phone and waits half a second. Then from the phone comes: S2: [Chinese] I changed my mind at the last minute. After hearing it, she lightly presses her lips together. Trying not to smile. Fails. The corner of her mouth still rises a little. Immediately enter the next shot. --- # SHOT 2 ## 00:05.5-00:13.5 ### "Nothing in her words, but her movements give everything away" The female lead stands up and opens the natural wood wardrobe door. The wardrobe door passes through the camera foreground. The entire frame is naturally filled by the warm brown wooden door. Use this real occlusion to complete a smooth Match Cut. When the wardrobe door opens again: The female lead is already dressed in her going-out look. Same character. Same room. Only a few minutes seem to have passed. The camera is now positioned to the front side of the wardrobe. Begins a very slight tracking to the right. The background music re-enters with a slightly clearer rhythm. The female lead stands in front of the mirror applying lip glaze. Very seriously presses her lips together once. Then looks at herself in the mirror. As if thinking: "Is this too serious?" She immediately takes a tissue and gently blots off a bit of the color. This small action must feel real. The music enters a very pleasant little drumbeat. While putting on earrings, she unconsciously follows the rhythm: Left shoulder lightly taps once; Right shoulder follows once; Body naturally sways left and right for two beats; Takes a step back; Then follows through with a very small, very relaxed half-turn. This is not formal dancing. More like a person whose mood suddenly improved, and without realizing it, her body starts moving to the music. The suede jacket, hair, and jeans produce real inertia. When she turns back, she just happens to see herself in the mirror. She also realizes she was a little too happy just now. She immediately stops. Can't help but smile a little. From the phone again comes S2: [Chinese] Two more minutes. While picking up her shoulder bag, the female lead replies: S1: [Chinese] I know, what's the rush. Her words are annoyed. But her movements clearly speed up. The contrast here is the joke and the sense of relationship. --- # SHOT 3 ## 00:13.5-00:20.0 ### "What he brought" The female lead goes out from the apartment door. The camera naturally follows behind her at an angle. Not a stabilizer-ad feel. Maintain a slight real follow-shot feel. She quickens her pace for two steps. Then deliberately slows down a little. As if suddenly realizing: She was acting too eager. She casually tidies her hair. Pretends to be very natural. Ahead in the corridor, the elevator: "Ding." The doors open. The camera does not immediately rush in. The female lead stands outside the elevator doors. She originally had a bit of a smile. But after seeing inside, Her footsteps slow. Her smile also gently freezes. Not fear. It is being struck by the memory of something. S2 still does not show his face. The frame only shows his hand reaching out from inside the elevator. In his hand he is holding: **An old silver portable CD player with obvious signs of use.** There is a slight scratch on one corner of the casing. A very old small blue sticker is attached to the body. It can be seen that this object has been kept for many years. The female lead stares at the CD player. Her previous relaxed state suddenly quiets down. The background music also lowers very naturally here. Leave a little corridor ambient sound. S1 says very softly: [Chinese] ...Why do you still have this? Do not cry. Do not immediately become sentimental. The male voice does not answer. Pause for half a second. The camera only very lightly moves forward a little. S2: [Chinese] Come down first, then I'll tell you. The female lead looks at the person in the elevator. Stops. In her eyes there is: Surprise; Familiarity; A little complexity; And very obvious curiosity. The elevator doors begin to close slowly. Just as the doors are about to cover the space between the two: End directly. No black screen explanation. No "to be continued." Let the audience naturally wonder: **What exactly is this CD player?** **What happened between them before?** --- # Dance Element The dance in this film must never become a large performance. The core is: **After the feelings can no longer be hidden, the body gives it away first.** Only use 4 very natural rhythmic actions: Shoulders for two beats; Light body sway; One step back; Small half-turn. The movements are like a real young woman alone, hearing a song she likes, and after her mood suddenly improves, moving along with it herself. Do not: Professional stage-performance feel; Large waves; High-difficulty dance; Long periods of dancing; Sexy gyration; Fixed TikTok hand dance. The function of the dance is: To show that she is actually very much looking forward to seeing this person. --- # Dialogue Design The entire film has only 5 lines. S2: "I'm downstairs." S1: "...Didn't you say you weren't coming?" S2: "I changed my mind at the last minute." S2: "Two more minutes." S1: "I know, what's the rush." Ending: S1: "...Why do you still have this?" S2: "Come down first, then I'll tell you." All dialogue must sound like normal young people chatting. No punchlines. No "dialogue-y" tone. No explaining the relationship. The audience, through: Pauses; Expressions; Speed of movement; The old object; guesses the relationship between the two. --- # Sound Design Background music: Light; Warm; With a bit of retro guitar and soft drum machine; About 95-105 BPM. Keep the volume low in the first half. When the female lead naturally does her little dance steps, the music becomes slightly clearer. After the elevator doors open: The music gradually lowers. Leaving: Corridor ambient noise; Shoe sole sounds; Elevator mechanical sounds; Slight breathing. When the old CD player appears, do not add an exaggerated "suspense Boom." The more natural the better. --- # Performance Curve 00:00-00:02: Ordinary state of being alone. 00:02-00:05.5: Surprise; Wanting to smile but holding it back. 00:05.5-00:10: Mood clearly improves. Movements begin to become lighter. 00:10-00:13.5: After being rushed, she talks tough, but is obviously anxious. 00:13.5-00:16: Looking forward to meeting. 00:16-00:20: After seeing the old object, her emotions suddenly quiet down. Not great sorrow or great joy. Leave a real sense of story. --- # Continuity Hard Constraints Same adult female lead. Same face. Same skin tone. Same body proportions. Same going-out outfit. The spatial logic from apartment to corridor is clear. S2 never shows his face. No sudden appearance of a complete male character. No second female lead. No character face drift. No sudden appearance of heavy makeup. No mansion-ization. No cyberpunk. No future UI. No supernatural phenomena. No sudden slow motion. No random orbiting shots. No hard costume changes. Costume change is only completed through a natural Match Cut fully occluded by the wardrobe door. No exaggerated short-drama crying scenes. No female lead constantly performing for the camera. Most of the time she should seem like: **She does not know someone is filming her.** --- # H3 Execution Priority ## First Priority Stable character identity; The first hook of "I'm downstairs"; The female lead's real subtle emotional changes; Wardrobe-occluded costume change; Natural little dance steps; The old CD player in the elevator at the end; "Why do you still have this?" These must hold up. ## Second Priority Real clothing materials; Natural follow shot; Curtain wind; Elevator mechanical movement; Hair and fabric inertia. ## Third Priority Specific scratches on the CD player; Small blue sticker; Small room decorations. If H3 execution complexity comes into conflict: Prioritize deleting small prop details. Do not sacrifice the character relationship and the final old-object foreshadowing. The final effect should be like a real slice of life accidentally captured: The first half makes people feel sweet and relaxed; The middle has a cute little contrast; The ending suddenly opens up a past story. The most natural question the audience should have after watching should be: **"What exactly was the relationship between these two before?"** Not: "How was this AI effect made?"
This clip was generated with MiniMax H3 by MiniMax. The prompt on this page is the exact text used to create it.
The example was generated with MiniMax H3 using: 40:23 · 0:20 · Dramatic · Cinematic · Realistic. Reusing these settings gives the closest match.
Copy the prompt as-is first, then change the subject while keeping the camera, motion, and lighting terms. Video models are sensitive to wording, so expect a few runs.
Yes. Prompts work best as starting points — swap scenes, adjust pacing terms, or change the aspect ratio to fit your project.