Afternoon by the Floor-to-Ceiling Window · Six Seconds from Chin-Rest to Lowered Gaze

Grok ImagineRealisticVerticalHealing6s
In warm afternoon light, an adult East Asian woman sits by a floor-to-ceiling window resting her chin on her hand and looking at the camera; 1.3 seconds in she slowly lifts her chin, then closes her eyes and lowers her head. Nearly the whole prompt is written around skin, hair, and light — keeping pores and peach fuzz, no skin-smoothing, with backlight turning the loose strands a translucent golden brown.
PROMPT · Video Prompt
Create with AI
A 6-second, 9:16 vertical, 4K ultra-photorealistic live-action video. A clearly adult East Asian woman sits beside a large floor-to-ceiling window, with a soft, natural oval face and realistic adult female facial proportions. Her deep brownish-black hair is pulled into a loose, languid low bun, with plenty of natural loose strands left at the forehead, temples, and around the ears; a few long fine strands at the back of the head drift out naturally and sway slightly with her movement. Clear, natural no-makeup makeup: natural brows, a soft apricot eyeshadow, realistic individual lashes, deep brown moist eyes, and coral-pink dewy lips. She wears delicate small silver drop earrings and a very fine collarbone necklace, and is dressed in a pale pink strapless bustier top. At the opening, her body is turned slightly to one side, one hand naturally resting against her cheek, head tilted slightly, quietly looking straight into the camera, the corners of her mouth carrying a faint, almost-there natural smile; After 1.3 seconds she slowly lifts her chin about 5–10 degrees, the hand resting on her cheek gradually relaxes, her slender fingers slide slowly down her cheek toward the jaw and chin, and her lips part naturally; After 2.6 seconds her head stays tilted slightly back, her neck and collarbone open up naturally, her index and middle fingers rest lightly beneath her chin, and her eyes gradually settle into a languid half-closed state; Starting at 3.8 seconds her eyes close very slowly, while her head gradually tilts downward, and the loose strands at her forehead and ears carry slight, realistic inertia from the movement; Finally her eyes are fully closed, forming a soft 3/4 profile, fingers lightly resting under her chin, holding a quiet pose of enjoying the afternoon sun with her head lowered, and this holds for 0.5 seconds. A 70–85mm portrait lens, close-up from the chest up, the camera roughly level with her eyes, almost entirely locked off for the whole shot, with only a 3%–5% extremely slow push-in and a slight handheld-breathing feel. The large window is on the left side, with warm afternoon sun slanting in from the front-left; realistic warm highlights appear on the bridge of the nose, nose tip, lips, shoulders, and collarbone, and the edges of her hair are lit by the sun into a translucent golden brown, against a minimalist warm-gray wall background. RAW real human skin, retaining fine pores, cheek peach fuzz, nose-wing texture, under-eye tone variation, slight skin unevenness, and real lip texture — no skin-smoothing, no plastic look, no AI look, with the person's features, hairstyle, outfit, and jewelry strictly locked throughout.
✍️ Editor’s Notes

This prompt breaks a single 'chin-resting' pose into five timed beats (open, 1.3s, 2.6s, 3.8s, and a 0.5s hold at the end), giving exactly one change per beat — lift chin, fingers slide down, eyes half-close, eyes close, hold — so the model isn't asked to cram too much motion into 6 seconds. The character description deliberately piles on anti-AI cues like 'clearly adult,' 'RAW real skin,' 'keep the pores,' and 'no skin-smoothing' — a common way right now to fight the 'plastic face' look in photoreal portrait prompts. The lighting names only one source (warm backlight from the left) but specifies five places it should land — nose bridge, nose tip, lips, shoulders, collarbone, hair edges — turning 'soft light' into something the model can actually execute instead of a vague adjective. The lens spec (70–85mm, a 3–5% push-in) is another case of translating restraint into a number instead of just writing 'keep the camera steady.'

Create with AI
Related Works
׋›
↑