REFERENCES
@Image1 (IMG_Clip_1.3.png) = START FRAME and character reference. The video begins exactly on this frame: the same woman lying on her stomach on a bed, propped on her forearms with her upper body slightly raised, wearing a light grey knit sweater, hydrogel under-eye patches (translucent lilac with small white floral motifs), thin gold necklaces with a small gold monogram pendant, long wavy chestnut-brown hair with warm highlights, olive-green eyes, freckles. White bedding, a striped pillow, an ornate tapestry-style headboard on the right, a paneled door and slanted ceiling behind her. No end frame: the clip ends wherever the action naturally lands.

CHARACTER LOCK
The same woman as @Image1 throughout, mid-20s. Identity must not drift: no face morphing, no change of eye color, hair color, necklaces or sweater. The under-eye patches stay in place, undeformed and unpeeled for the entire clip, even when she moves her hair. No new accessories appear.

VOICE
American English, female, mid-20s, slightly low, relaxed, a little dry and amused. Casual, deadpan, like talking to a friend on a lazy day, genuinely incredulous. Natural, continuous pace. The voice starts at the very first frame of the clip, with no initial silence. Not a commercial voice, not theatrical. Close-mic phone quality with slight room tone. Lips sync exactly to the words.

SETTING
Her bedroom in soft natural daylight: cozy, lived-in, unmade white bed, striped pillow, warm neutral walls. Quiet and private. No other people.

FORMAT
Duration 5 seconds. Aspect ratio 4:3 (1440x1080). One continuous shot, no cuts. Natural motion, 24 fps.

SHOT BREAKDOWN
Handheld phone selfie at low bed level, camera looking slightly up at her, same framing as @Image1 at the start. She speaks while she moves: all the physical actions below happen WHILE she is talking. There is motion and speech from the very first frame.

0.0-0.4 s: She lets her whole body drop forward and straight down onto the bed, a relaxed collapse: forearms slide forward, chest and shoulders sink into the white bedding with a soft thump. As she drops, a few strands of her hair fall across her forehead and one cheek (never covering her mouth). The camera gives a small natural bounce and ends lower and closer to her face (framing tightens about 15%). She is already talking, casual and a bit tired: "Okay," said mid-drop, eyes on the lens.

0.4-1.9 s: While continuing to speak, she lifts one hand, brushes the strands off her face and tucks them behind her ear in one quick natural gesture, eyes on the lens the whole time. She says, relaxed and incredulous: "it still blows my mind". Stress on "blows my mind". Eyebrows rise above the patches, eyes widen a little. Her hand returns to the bed as the line ends.

1.9-2.1 s: Very short beat of silence. A small slow head shake of disbelief, eyebrows still raised.

2.1-3.7 s: She says, slower and deadpan: "that nobody can tell I'm AI." Stress on "nobody" and on "I'm AI". On "I'm AI" she lifts one hand from the bed, points a finger at herself, then lets it drop back down.

3.7-5.0 s: She stops talking, tilts her head and rests her cheek on her forearm/pillow, lying a little more, holds a calm direct stare into the lens, one slow blink, a faint knowing half-smile. Her face stays close to the camera.

CAMERA
Handheld front phone camera at bed level, about 24mm equivalent with slight natural wide-angle distortion. A natural bounce and downward move at 0.0-0.4 s from the drop, then gentle handheld sway and breathing movement with a slow push-in of 5-8% for the rest of the clip. Shallow natural phone depth of field: her face sharp, room softly blurred. No whip pans, no cuts, no sudden zooms.

LIGHT AND LOOK
Soft natural daylight from the left, gentle falloff, neutral warm tones, realistic and stable catchlights in both eyes. Authentic iPhone front-camera video look: mild sensor noise, slight softness at the edges, natural dynamic range, no HDR halos, no over-sharpening.

AUDIO
Her voice is present from 0.0 s, close and clear, over a soft thump and rustle of bedding and knit fabric as she drops and a faint brush of hair as she tucks it. Soft room tone. No music, no sound effects, no voiceover.

GLOBAL REALISM
Visible pores, fine facial hair, natural freckles, real eye moisture, natural blinks every 2-4 seconds, subtle cheek and jaw movement when talking, soft breathing in the shoulders. Individual hair strands fall, slide and tuck behind the ear naturally, with no clipping through the face or patches. The hydrogel patches look glossy and realistic, adhered to her skin. Looks like a real phone video recorded on a bed, not a rendered avatar.

NEGATIVE
No silent start, no static start, no frozen pose, no face morphing or identity drift, no changing hair color, necklaces or sweater, no hair covering the mouth while she speaks, no hair merging into the face, no peeling, sliding or disappearing patches, no plastic or waxy skin, no beauty filter, no exaggerated mouth movements, no lip-sync mismatch, no extra fingers, no distorted hand when tucking hair or pointing, no phone visible in frame, no text, no subtitles, no watermark, no logos, no cuts, no cartoon or illustration look, no second person.