Cheongsam Beauty in Warm Ancient Style AI Image Prompt

9:16 vertical. 0-3s: Scene: ancient-style interior, warm soft light on floor, a tall and curvy beauty in a long off-white cheongsam with golden peony pattern standing by the window. Camera movement: slowly push in from her back, focus on her profile, face rotation strictly within 30°. Action: she raises both hands simultaneously, slowly opens them outward from chest, fingers gently brushing her shoulders and upper arms, waist gently leaning back forming a smooth curve, face kept as much as possible toward camera. Sound: pipa as main melody opening, flute as accompaniment, wind blowing curtains rustling, faint humming. Mood: gentle and stretched, warm light outlines the cheongsam silhouette, golden peony pattern faintly visible in soft light. 3-7s: Scene: camera orbits around her, showing profile and details of golden peony pattern on off-white cheongsam. Camera movement: small smooth orbit, avoid large angle turns to prevent face distortion. Action: slightly turn, skirt gently flutters; right hand slides from shoulder along collarbone to opposite arm, left hand naturally on hip, face always at small angle, when facing camera slightly raise chin, gentle restrained smile. Sound: pipa gradually climaxes, flute supports, humming slightly louder, ambient sound fades. Mood: elegant and stretched, restrained charm, pattern shows light-dark changes with movement. 7-10s: Scene: eye-level close-up of her eyes, eye contact with lens, no low-angle perspective. Camera movement: steady push in, eye-level, no large up/down tilt. Action: slight head tilt, eyes blink slowly once, mouth corners slightly raised, right hand gently fans near cheek, facial features remain stable throughout. Sound: music gradually softens, leaving only pipa and flute remains, humming fades. Mood: charming but not frivolous, languid and gentle, image fades to dark. Locking instruction: only lock facial features, bone structure and body shape; hairstyle, clothing, scene strictly follow script, never replicate original image; keep as the same person throughout, no facial drift, face distortion or facial deformation. Negative constraints: no copying original styling, no highly similar images, no excessive face angles, no distorted facial features.