Vlog intimiste d'un selfie dans un café urbain

Generate a short vertical video (9:16) of a woman taking a selfie with her phone in a Starbucks café setting. The video mimics a genuine front-camera phone recording with handheld, non-stabilized movements.

Character (locked consistency): Female with refined makeup — coral-pink lips, soft eyeliner, wispy bangs framing her face. Maintain her facial features, hairstyle, clothing, and accessories consistently throughout.

Camera & Editing Logic: The camera is held by the subject throughout. Frames shift between extreme close-ups (forehead, eyes, nose, lips filling the frame) and mid-chest framing. Movement is jerky and unselfconscious — the phone suddenly pulls away from her face, causing the image to drop briefly to chest/waist/thigh level with slight blur and frame shake, then is abruptly brought back close to her face. No dolly or gimbal smoothing; every movement feels authentically handheld and random.

Expression & Motion: Transitions between coy and flirtatious — starts with downcast eyes and closed mouth, then looks directly at the camera with subtle head tilt and gentle blink. A barely-there smile plays across her lips. Include one extreme close-up of just her lower face and mouth. Throughout, capture natural hair flips, slight head turns, and breathing movements.

Audio: Layer an original R&B-pop track (BPM 100–108) with clear rhythm, breathy female vocals, deep bass, and sharp electronic drum hits. Sync camera pickups and pullbacks to beat strikes. Add natural sound layer: rustling fabric, hair brushing sounds, subtle nearby breathing, light laugh, and ambient café ambience.

Visual Style (Hard Requirements): Replicate TikTok/Instagram real-time selfie aesthetic — not polished cinematography. Apply soft beauty-filter beautification with skin brightening, but retain fine grain texture, subtle color noise, and visible compression artifacts in midtones and shadows. Include motion blur and brief focus shifts during rapid camera movements. Preserve focus breathing, rolling-shutter wobble, exposure pumping, and slight desaturation in blacks. Target 720p compression appearance. Absolutely no watermarks, no uniform sharpness, no HDR, no studio lighting, no gimbal-smooth trajectories, no noise reduction, no high-contrast grading.

Pacing: No dialogue, no lip-sync. Open on a mid-chest frame, cut abruptly to extreme close-up of face, pull back to show hair and upper body, then hard-cut back to tight facial close-up. End holding on extreme close-up with sustained flirtatious eye contact and visible breath.