클립 1 / 14
생성 프롬프트 · 10782자
[MATERIAL ROLES] is a reference sheet of ONE SINGLE WOMAN — Member A — photographed from four angles side by side plus a face close-up inset. It shows one person, NOT four people and NOT five people. is the same kind of sheet for ONE SINGLE WOMAN, Member B. is the same for ONE SINGLE WOMAN, Member C. is the same for ONE SINGLE WOMAN, Member D. is the same for ONE SINGLE WOMAN, Member E. Each sheet defines that one woman's face, hair, makeup and outfit only. Preserve them exactly and retain true pores and skin texture. Do not use their backgrounds, poses, expressions or lighting. Never blend faces — each woman keeps her own independent features and outfit in every shot. Member A has cherry-red hair in a high ponytail. Member B has dusty rose waves and an ivory outfit. Member C has an inky blue-black bob. Member D has long platinum hair and a camouflage vest. Member E has plum violet hair with a fringe and a burgundy velour outfit. defines the white studio location. defines the flower-filled subway carriage. defines the drained swimming pool. defines the city backstreet. Use only the architecture, materials, colour and light of each; do not add any person from these images. defines the complete music track and the lead vocal for the entire video — its tempo, its structure and every word sung. Do not describe, alter, replace or add to this music, and add no other sound. [HEADCOUNT — ABSOLUTE RULE, APPLIES TO EVERY SINGLE FRAME] There are exactly FIVE women in this entire video: Member A, Member B, Member C, Member D and Member E. One of each. No more. Each member appears exactly ONCE in any given frame. Never duplicate, clone, mirror, twin or repeat a member. There is never a second Member A, never a second red ponytail, never a second Member B, never a second Member C, never a second Member D, never a second Member E — not beside her, not behind her, not in the background, not out of focus, not in a reflection, not through a doorway, not at the edge of frame. No sixth person of any kind ever appears. No backup dancers, no crew, no bystanders, no silhouettes, no partial bodies entering frame. When the camera is in close-up on one member, the other four are simply OUTSIDE the frame. They have not disappeared and they must not be duplicated into the background to fill the shot. If a shot would ever contain six or more figures, remove the extra figure and keep only the five named members. [LOGLINE] A five-member female performance group dances one continuous routine across four locations, cut to the beat of , ending in an eight-second rapid intercut through all four. [GLOBAL SETUP] 1. Cast: exactly five performers, instantly told apart by hair colour — red, rose, blue-black, platinum, violet — subject at all times to the HEADCOUNT rule above. 2. Style: high-end cinematic music video, shot on an ARRI Alexa with anamorphic lenses. Atmospheric haze, cinematic colour grade, deep blacks, controlled highlights, fine film grain, shallow depth of field. Expensive and polished — never documentary, never raw, never run-down. 3. Timing: every dance hit, every formation change, every camera move and every cut lands exactly on the beats of . The beat falls every 0.469 seconds and the bar every 1.875 seconds. 4. Lipsync: only the member named in each stage sings, and her mouth is synchronised to the vocal in for that stage. The others keep dancing and never mouth the words. 5. Performance: full-out fast choreography — sharp arm hits, hard chest pops, floor-slide footwork, synchronised head snaps. Confident, sharp, slightly cold. 6. Cuts: the routine is ONE unbroken choreography from beginning to end. When the location cuts, the dancers are mid-movement and the movement continues seamlessly into the next location on the same beat — same body position, same direction, same energy. Prohibit rigid cutting, prohibit anyone appearing out of thin air. 7. Reflections: floors may be glossy but no reflection ever resolves into a recognisable extra figure. A reflection is never counted as a person and never becomes one. 8. Never: no subtitles, no on-screen text, no logo, no watermark. No particles, no lens flares beyond anamorphic streaks, no AI gloss. [TIMESTAMP STORYBOARD] Stage 1 — 0.000 to 11.250 seconds. Location . ONE CONTINUOUS TAKE, no cuts. Five women in this stage, never more. All five sing together, all five mouths synchronised to . 0.000: overhead top shot. Exactly five women on the white floor — one at the centre and four around her in a radial star, heads down. On the first beat they lift their heads and begin the routine. 1.875: camera dives vertically to a dead-centre symmetrical eye-level wide. The five spread into a single row facing camera — count them: five, no more — dancing full-out in unison. 3.750: the camera pushes in until Member A ALONE fills the frame in close-up. The other four are outside the frame entirely; nobody stands behind her and no second red ponytail appears anywhere in the shot. 4.688: the camera reframes to Member D ALONE in close-up, the other four outside the frame. 5.625: reframes to Member B ALONE in close-up, the other four outside the frame. 6.563: reframes to Member C ALONE in close-up, the other four outside the frame. 7.500: reframes to Member E ALONE in close-up, the other four outside the frame. 8.438: the camera pulls back out to a full-group wide. Exactly five women are now visible together, one of each. 9.375: the bass drops out completely for one bar. All five freeze mid-move, absolutely still, camera slowly craning up and back. Member C raises one open hand toward the lens and sweeps it across, covering the frame. End state: her hand fills the frame. Stage 2 — 11.250 to 17.813 seconds. Location . ONE CONTINUOUS TAKE, no cuts. Five women in this stage, never more. Member B and Member C are in front and sing; Members A, D and E dance behind them and do not sing. 11.250: the hand pulls away and the frame opens inside the flower-filled carriage — the same five women already mid-movement, continuing the exact motion they froze on. Camera dead-centre down the aisle in one-point perspective. Two in front, three behind: five total. 11.250 to 15.000: the camera pushes slowly forward down the aisle as B and C dance toward the lens through the flowers, with A, D and E behind them among the blooms. Petals lift in their wake. No further figures are visible further down the carriage. 15.000: camera settles into a two-shot close-up of B and C only. The other three are outside the frame at this moment and are not duplicated into the background. 17.813: the bass strips away — Member D turns fully at the edge of frame and the camera swings with her into motion blur. End state: horizontal motion blur, movement travelling left to right. Stage 3 — 17.813 to 22.500 seconds. Location . ONE CONTINUOUS TAKE, no cuts. Five women in this stage, never more. Member D sings alone; the other four dance and do not sing. 17.813: the blur resolves inside the drained pool travelling the same direction, the same five women arriving in the basin mid-move. Camera high overhead looking straight down the long axis — exactly five bodies visible on the tile from above, no more. 17.813 to 20.625: the overhead descends as four of them rotate below in a slow kaleidoscope and Member D moves out to the centre for a featured solo, sharp isolations landing on the beat. Four rotating plus one centre equals five. 20.625: the camera drops to a low angle between two shafts of skylight, following Member D as she moves through the beams. The other four rise into a group behind her — four, not five. End state: Member D centred with four behind her, five in total, all surging forward on the last beat. Stage 4 — 22.500 to 30.040 seconds. RAPID INTERCUT — a hard cut every half bar, eight shots, each location appearing twice. Five women in every one of these eight shots, never more. Member A leads and sings from 22.500 with Member E answering beside her; from 26.250 all five sing together. Across all eight shots the choreography runs unbroken — the same five women hold the same continuous routine and the same forward-driving direction, and only the location changes on the cut. Each cut is a hard white flash frame landing exactly on the half bar. The five are the same five in every shot; none of them is ever doubled when the location changes. 22.500: , the city backstreet. Speed-ramped low angle along the wet asphalt, the five moving toward camera spread across the frame, Member A at the front singing with Member E at her shoulder. One red ponytail and one violet head in this shot, not two of either. 23.438: , the flower carriage. The same five continue the same movement down the aisle, petals bursting up around them. 24.375: , the pool. The same five continue up the sloping basin floor, a hard shaft of skylight crossing their faces. 25.313: , the backstreet. Tight tracking two-shot on Member A and Member E, their faces sharp side by side. Only these two are in this frame — the other three are outside it, and there is no second red ponytail and no second violet head behind them. 26.250: , the white studio, on the beat the full bass returns. Wide dead-centre symmetrical shot, the five settling into a V formation with Member A at the point, hitting the hardest unison of the piece, all five singing. 27.188: , the flower carriage. Overhead top shot looking straight down on exactly five women dancing among the blooms. 28.125: , the pool. Low angle from the basin floor, five figures against the skylight beams, arms rising together — five silhouettes, not six. 29.063: , the white studio. Member A raises her right arm and the other four settle into four different sculptural poses around her, all five coming to stillness. The camera cranes up and pushes in, the composition settling to match the opening frame of Stage 1 for a clean loop. End state: a still tableau of exactly five women, all five faces sharp. [MAINTAIN CONSISTENCY] The same five women — one Member A, one Member B, one Member C, one Member D, one Member E — in all four locations and in every one of the eight intercut shots. The headcount is five in every single frame of the video: never six, never seven, never a duplicate of anyone, never a background figure, never a reflected figure counted as a person. The music from runs unbroken across every cut with no gap and no tempo change. No subtitles, no on-screen text, no logo, no watermark. No particles, no AI gloss.