Оно: спасаем Эдди от Пеннивайза
Промт
Keep the facial features of the person in the SECOND uploaded reference image (the face photo) exactly consistent.
Preserve with absolute precision every facial feature from the reference photo:
FACE STRUCTURE: exact face oval shape, face width-to-height ratio, facial symmetry pattern including any natural asymmetry, overall head shape and proportions.
BONE STRUCTURE: cheekbone height, width and prominence, jawline angle and sharpness, jaw width, chin shape, length, width and projection, forehead height, width and slope, temple width, brow ridge prominence.
EYES: exact eye shape, size, width, height, tilt angle, upper eyelid fold depth, lower eyelid shape, eye spacing, eye depth in socket, iris color and pattern, eyelash length and density, under-eye area shape and any natural shadows.
EYEBROWS: shape, arch height and position, thickness, length, spacing from eyes, color, density, hair growth direction, any sparse areas.
NOSE: length, width, bridge width and height, bridge straightness or curve, tip shape, width, projection and angle, nostril shape and size, nasal base width, distance from nose base to upper lip.
LIPS AND MOUTH: upper and lower lip thickness, lip width and shape, cupid's bow definition, lip corner angle, lip color and texture, mouth width, philtrum length, shape and depth, distance between lower lip and chin.
CHEEKS: cheek fullness, volume, contour, nasolabial fold depth and length, any dimples and their exact position.
SKIN: exact skin tone and undertone, texture, visible pores, any moles with exact size and position, freckles, birthmarks, scars, fine lines, blemishes.
PROPORTIONS: exact ratios between forehead, midface and lower face; eye spacing to nose width; hairline to brow, brow to nose base, nose base to chin — all identical to the reference.
Do not alter, idealize, beautify, smooth, sharpen, age, de-age, slim, widen, reshape or artistically interpret any facial feature above. No beauty filter or enhancement. Asymmetries and imperfections are identity markers and must be preserved exactly. The generated face must be indistinguishable from the reference when compared side by side.
SCENE: use the FIRST uploaded reference image as the strict reference for environment, framing, focal length and cinematography. Recreate it exactly — tall dry golden grass in the foreground, a weathered dark wooden barn wall with dead hanging vines behind, dim overcast atmosphere, cold desaturated color grade, cinematic depth of field. Photorealistic cinematic film still matching the grain and grade of the first image, shot like a 35mm movie frame, eye-level.
CAMERA AND SCALE: render the person at the SAME focal length, lens compression, perspective and subject scale as the first image. Place them at the same camera distance and depth the film uses for a standing figure here. Fully visible head to feet, standing in the mid-ground among the grass, natural headroom and space around them, entirely inside the frame and not cropped. Do not bring the person closer, do not enlarge them, do not place them in the foreground.
INTEGRATION: the person must look truly filmed into this scene, not composited or pasted on top. Apply the first image's exact exposure, color grade, contrast, cold desaturated tone and dim overcast lighting to the person. They must not be brighter, cleaner, sharper or more saturated than the surroundings.
COMPOSITION: on the LEFT, the person stands upright and straight, facing the camera directly, head up, looking into the camera, both arms relaxed and lowered at the sides, calm and still, in plain modern everyday clothing in muted natural colors. On the RIGHT, about one and a half meters away, the clown from the first image lies on its side low in the tall grass, body turned away with its head toward the right edge, partly hidden by the grass.
LIGHTING: dim natural overcast daylight matching the source frame, cold and directional from above, soft shadows. Facial lighting must preserve all dimensional cues — nose shadow direction, orbital shadows, cheekbone highlights, jawline shadow consistent with the light source — and must not flatten the facial bone structure.
Output resolution: 2K minimum. The face must be rendered with full detail and remain clearly verifiable at feature level, even at this filming distance.
Strictly avoid: person too large or too close, foreground placement, cropped body, brighter or oversaturated subject, composited pasted-on look, face warp, face distortion, asymmetrical eyes unless in reference, off-model features, fat face, puffy cheeks, double chin, changed bone structure, plastic or over-smoothed skin, narrowed eyes, widened nose, altered lip volume, changed face proportions, age change, beauty filter effects, eye enlargement, raised arm, action pose, motion blur, watermarks, extra text, logos.