Create a native 15-second, vertical 9:16 handheld iPhone video filmed by a person standing in a crowd at a tech expo booth. One continuous shot, no cuts. The uploaded image is the exact first frame of the video. REFERENCE IMAGE LOCK The android girl in the center of the image is NOVA. Keep her 100 percent identical to the image in every frame: the same face, facial features and proportions, the same light hazel eyes, brows, lips and skin tone, the same jet black hair with center part, the same teal crop T-shirt, the same high-waisted black bike shorts with silver buttons, the same body shape, and the same exposed mechanical waist band with silver pistons, springs and servos. Do not redesign, beautify, age or slim her. Keep the same booth, crowd, ceiling, light, colors and camera angle as in the image. The spectator hand holding sunglasses on the LEFT edge lowers out of frame within 0.5 seconds and never returns. Nova wears no sunglasses in the entire video. CORE RULES One continuous handheld shot. One main character: NOVA. Nova stays on the same spot for the whole video: she never walks, never turns around, never spins. From 00:04.0 there is an empty circle of at least 3 meters around her; nobody is within 3 meters of her while she moves. She never touches anyone. Exactly two people fall, and they fall from fright at the loud spark crack, while Nova is still standing frozen and not moving. Nobody falls while Nova is striking. Everyone else stays standing for the whole video. Nobody is injured. No text, captions or overlays on screen. All people are adults aged 18 to 30. CHARACTER NOVA looks exactly like the image: a female android who looks like a real young woman, apparently 23 years old, slim, very narrow waist, full, large, natural-looking bust, long athletic legs, realistic body proportions. Straight glossy jet black hair, center part. Fitted teal short-sleeve crop T-shirt, high-waisted black bike shorts with a row of small silver buttons, black sneakers. THE ROBOT WAIST: exactly as in the image, a fully exposed band between top and shorts with brushed silver steel actuators, vertical pistons with coil springs, small servo motors, black cables and aluminium plates. Always visible, never changes shape. In fight mode tiny blue status lights blink inside it. FACE Identical to the image in every frame. Normal mode: sly closed-lip half smile, eyes open and on the camera; she never looks down. Fight mode: blank, emotionless machine face, eyes locked on the camera, lips closed. Only upper teeth if teeth show. Her face is always sharp and never motion blurred. Her hair never covers her face. VOICE Soft, calm, slightly synthetic female voice with a small digital edge. One line only, at the end. CAMERA Held by a spectator about 2 meters in front of Nova at chest height, pointing slightly down, as in the image. At 00:04.0 the camera person steps back about 1.5 meters once, so Nova is visible from head to feet inside the empty circle with the crowd around, and stays there to the end. A young woman in a denim jacket holding a black tablet is on the RIGHT edge at the start. CAMERA HAND The phone holder is never visible. HANDHELD Natural small shake in the calm part, a strong jolt at the spark crack, then a slightly nervous shake that keeps Nova centered. No zoom. LOOK Plain iPhone video at 60 fps, default phone processing, light sharpening, natural noise, no filters, no color grading. Real documentary feel. LIGHTING Even overhead light from rows of round LED ceiling lamps in a large exhibition hall. Neutral cool white, soft shadows, her face always well lit. VISUAL MATCH The image must look exactly like the uploaded image: a phone video from inside a dense crowd at a tech expo, neutral grey and white hall, open steel truss ceiling with many round lamps, the teal top as the only strong color accent, black T-shirts with blue lanyards on the crowd, blue panels behind, everyday iPhone sharpness, no overlays. SETTING The same robotics booth inside a huge modern exhibition hall during the day. BACKGROUND As in the image: about 20 young adults with blue lanyards around Nova, several filming with phones, a metal cart with robot parts behind her, white booth walls, a dark blue backlit poster on the LEFT, a blue flag, steel truss ceiling. T-shirts are plain black with no logos or text. One distant movement: a person walking far behind the booth. MOTION Normal mode: human-like but slightly too precise. Fight mode is a malfunctioning robot running a martial arts program: every move is a sharp mechanical jerk that snaps into a pose in 0.15 seconds and then stops dead and holds completely still for 0.4 seconds before the next move, like a robot hitting fixed positions. No flowing dance moves, no swaying, no circling arms. Each snap has a servo click and a whoosh. Hair swings on each snap and settles during each hold. DEFECT PREVENTION Hands: Nova's hands are either tight fists with the thumb outside or flat open palms with fingers together, five fingers each, never merging. Every strike goes into empty air in front of her or to her side, never toward a person and never toward the lens. The woman on the RIGHT holds the tablet with both hands until she steps out of frame at 00:03.8. The camera hand is never visible. Face: identical and sharp in every frame. Her head never jerks or snaps; she always faces the camera, head turns under 20 degrees. Hair never covers her face. Body and legs: exactly two arms and two legs. Her feet stay on the same spot all video. Every kick uses one leg while the other foot stays flat on the floor carrying her weight; knee up first, then a straight extension forward no higher than waist height, then the foot back down to the same place. No jumps, no spins, no turning her back, no walking. Knees bend only forward over the toes. People: from 00:04.0 nobody stan
This clip was generated with Seedance 2.5 by ByteDance. The prompt on this page is the exact text used to create it.
The example was generated with Seedance 2.5 using: 9:16 · 0:15 · Cinematic · Sci-Fi · 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.