Невысказанное прощание на платформе метро

Исходный промпт (без изменений)

[SPEC] 12 seconds, 9:16 vertical, 24fps, 1080p, realistic urban relationship short, one continuous handheld follow shot, no dialogue.

[CHARACTER LOCK]
Woman: a clearly adult 25-year-old East Asian woman with distinctly East Asian features, delicate face, long straight hair, polished daytime makeup; a light camel trench coat, white shirt, black tailored trousers, holding a half-finished cup of coffee.
Man: a clearly adult 27-year-old East Asian man with distinctly East Asian features, short black hair, clean brows and eyes; a dark grey knit sweater under a black jacket, a commuter backpack.
Both keep the same face, outfit, hairstyle and body proportions for the entire clip.

[SETTING & LIGHT] A city metro platform at evening rush hour in China: white tiled walls, platform screen doors along the track, rows of cool white ceiling tubes, reflective floor. The train has not arrived yet; the platform is not crowded, with a few blurred waiting passengers in the distance.

[SHOT LIST]
0-2.4s: medium shot, the two stand one step apart, neither speaking. She looks down at her phone, he keeps his hands in his pockets, gaze toward the tracks.
2.4-4.8s: she looks up as if about to speak, her lips move and stop; he notices and turns to her.
4.8-7.2s: she offers him the coffee. He pauses, then takes it. Their fingertips touch briefly and part.
7.2-9.6s: the wind of an approaching train is heard in the distance; her hair and coat hem lift in the airflow. She takes half a step back and puts both hands in her pockets.
9.6-12s: the train arrives, light sweeping quickly across her face. She turns toward the doors without looking back; he stays where he is, the coffee still raised. The doors close and the frame holds on him.

[TEXTURE] Real handheld follow shot, hard light from cool white tubes, shallow depth of field, slight noise; keep the crowd defocused, no sharp faces of passers-by.

[SOUND] Ambience only: an unintelligible station announcement, the wind and brakes of the arriving train, footsteps, fabric rustling. No music, no dialogue.

[NEGATIVE PROMPT] kissing, hugging, melodramatic crying, visible lip-sync of dialogue, sharp faces of passers-by, Western/Caucasian bone structure, non-Asian face, deformed background crowd, face swap, outfit change, deformed fingers, garbled station signage, music, subtitles, watermark.