楼梯平台上的日本女性写实人像
楼梯平台上的日本女性写实人像，适合短视频封面、AI生图、AI视频分镜和内容创作参考。模型：ChatGPT。
AI提示词短视频灵感
示例图
图片来自公开来源，仅作提示词效果参考；正式商用请按来源授权处理。
提示词正文
Subject: The finished image is vertical 9:16, a single independent photo. Screen width 100%, height 100%, center of the face horizontally 49–51%, the line connecting the eyes located at the upper end 27–32%, head top margin 3–7%, and the overall visual center of the person fixed at 45–55% horizontally. The head and hair are at least 4% from the left and right edges, the face height is approximately 18–25% of the screen height, and from the top of the head to the lowest part of the visible lower body is contained within 88–96% of the screen height. Only one protagonist, no left-right reversal of the character, no replacement with others, no clothing changes, and no background reorganization. Priority is given to the positional relationships visible in the attached image, and invisible elements outside the frame are not supplemented by speculation. Retain the natural resolution of a smartphone shot, slight perspective differences from close range, and the sense of imperfection in spontaneous composition. On an inorganic gray staircase landing, photograph only one fictional adult Japanese woman, who sits with her body close to the immediate front of the camera. The woman looks like a young adult over 20 years old, with a calm sense of intimacy and a natural aura of daily life. The confirmed screen range includes the head, face, neck, shoulders, chest, part of the abdomen, arms, and the top of the jeans; the full legs, shoes, and indoor panoramic view are all outside the frame and not produced. Do not rejuvenate the subject's age to a minor, maintain adult-like facial features, body proportions, and a sense of composure. Do not replace the character's ethnic appearance with other characteristics, making it a natural look that can be read as a Japanese woman. Do not add faces, bodies, shadows, or reflections other than the person themselves. The connection of the face, neck, shoulders, and torso is anatomically natural, and even if slim, do not unnaturally eliminate bones or joints. The range of visible skin, clothing coverage, and boundaries with the background remain consistent. Face/Makeup: Maintain a small oval face, a gentle forehead, naturally rounded cheeks, and a slender jawline. The skin is a bright neutral beige, with a very faint flush on the cheeks and tip of the nose. Eyebrows are a natural parallel shape in dark brown, eyes are deep brown, the upper eyelid line is restrained, and no strong lines are drawn on the lower eyelid. The bridge of the nose is thin and straight, and the lips are a light rose beige, lightly closed. Avoid heavy application, strong contour correction, excessive eyelashes, or porcelain skin. Do not make the face perfectly symmetrical left and right, retaining minimal natural differences in eye opening, eyebrow height, cheek shadows, and corners of the mouth. Skin tone is continuous with the neck, shoulders, and arms, without separately brightening the face or changing the color temperature. Avoid vinyl skin that eliminates pores, excessive cheekbones, proliferation of double eyelid lines, or fusion of teeth or lips. Eye reflections are only a small catchlight consistent with the main light source, without adding patterns or glows to the iris. Hair: A thick, deep dark brown bob reaching shoulder length. Bangs are thin, non-uniform strands hanging toward the forehead, with uneven volumes on the right and left sides. The ends of the hair slightly flare inward and outward, with fine wisps of hair remaining near the right cheek. Suppress the puffiness at the roots, and place a subtle sheen only on the light-receiving top surface. Do not change to long hair, strong curls, bright hair colors, or overly neat salon styles. Hair is not depicted and filled as single strands, but as an overlay of hair bundles of different thicknesses, clumps, shadows, and a small amount of stray hairs. The boundary between the head and scalp is natural, and bangs do not intrude into the eyebrows or pupils. Areas where ears are not visible are not forced to be exposed, and invisible structures of the back of the hair or hair knots are not created. Do not create cutout white edges or blending at the boundary with the background; only restrained light transmission and sheen are placed on the light source side. Expression/Gaze: A quiet gaze looking directly at the camera, and a gentle expression with the corners of the mouth hardly raised. The gaze does not strike a provocative pose, waiting softly for the other person's reaction at close range. Do not open the eyes wide, retaining natural micro-differences between the left and right eyes, eyelid weight, and the relaxed state of the cheeks. Do not replace with a toothy smile, a wink, exaggerated surprise, or a fake smile. Pupil direction, upper eyelids, lower eyelids, eyebrows, cheeks, and corners of the mouth are integrated as a single emotional consistency. Avoid situations where the left and right pupils face different directions, focus is off, or only the gaze deviates from the direction the face is facing. Do not exaggerate the expression into an advertising-style fake smile, retaining the subtle tension and relaxation of a natural distance from the photographer. Do not add emotions or stories to the expression that cannot be confirmed in the image. Pose/Fingers/Contact: A close-up pose with the upper body leaning forward and both shoulders slightly tucked in. One arm extends downward into the frame, and the other arm is placed near the body. Fingertips are only partially visible at the lower end; do not speculate or supplement invisible fingers or wrists. The center of gravity of the body is low, and due to being close to the camera, a perspective difference is created between the chest and the face. Do not unnaturally lengthen the neck, do not make shoulder heights perfectly symmetrical, and do not create the sitting surface or leg shapes. The axes of the head, neck, shoulders, thorax, and waist are continuous and consistent with gravity and contact points. Do not add extra elbows or wrists, and coordinate the thickness and joint positions of the visible arms. Maintain the way fingers are cut off when they are only partially visible, without adding extra fingers to the hands outside the finished frame. Add moderate compression, shadows, and folds at the contact points of clothing or the background, so that the body does not appear to float from the surface. Clothing/Accessories: A cream-colored ribbed crop top with thin spaghetti straps. Maintain fine vertical textures, a shallow curve at the chest, and a stretch feel that fits the body, without emphasizing transparency or excessive exposure. Only the upper part of blue jeans is visible at the bottom; metal buttons or the full shape of pockets are outside the frame. Do not add accessories, hats, watches, flashy nails, or bags. Do not change the color, material, strap width, or length of the clothes. Fabric stitching, openings, boundaries at the shoulders or chest, overlaps, and stretching directions are consistent with the pose. Do not make the clothes a painted surface attached to the body; distinguish fabric thickness, looseness, stretching, and fine folds. Do not add brands, patterns, text, decorations, or hardware not present in the frame. When small items are only partially visible, maintain only the visible shape and do not create the back or the overall form. Do not make changes that increase exposure or emphasize sexiness. Location/Time/Background: The background is a gray concrete or stone staircase landing. There are dark horizontal and diagonal boundaries at the back left, and bright wall surfaces and steps on the right; beds or furniture suggestive of hotel rooms are not generated as they are not in the frame. Do not change to a luxury interior, cafe, or outdoors. The amount of background information is lower than the character, while retaining seams, slight scratches, and light gradients of inorganic surfaces. The time period is limited only to the range readable from the light in the frame; do not add new scenery determining morning, noon, or night. Maintain the overlap of foreground, middle ground, and background, without causing the background to unnaturally bend along the character's outline. Background information is restrained relative to the character but is not replaced with a monochromatic blur. Do not generate windows, doors, signs, furniture, pedestrians, vehicles, scenery, or reflections that cannot be confirmed within the frame. Composition/Camera: A wide-angle feel like a close-range smartphone camera. The camera is aimed diagonally downward from a position slightly higher than the woman's face, naturally showing the distance difference between the face and the upper body. Avoid excessive ultra-wide-angle distortion, but do not eliminate the slight perspective difference of the arms or chest characteristic of close-up shots. With the face and eyes as the main focus, the background is not completely blurred but softened to a degree where shapes are recognizable. Do not change the person's occupancy on the screen, face position, left and right margins, or the way the bottom is cut off. Fix the camera height, tilt angle, and distance from the subject; do not change to a frontal ID photo style or a long-distance full-body shot. The sense of focal length remains at a smartphone standard to slightly wide-angle level, avoiding ultra-wide angles where the background straight lines and face proportions break down. The depth of field is such that the face and main clothing are readable, and the background is gently softened. Do not create an excessive loss of outline in portrait mode. Light Source: The main light source is strong natural light entering from the upper right of the frame, illuminating the right shoulder, cheek, bridge of the nose, and the right side of the top. Soft shadows are retained on the left cheek and the inside of the hair, without uniformly lifting the dark areas. Fill light appears only as weak reflections from surrounding walls. Do not change the light source to a frontal flash, sunset, neon, or studio lighting; unify the shadow direction and highlights. The direction of light falling on the skin, hair, fabric, and background is all consistent. Place natural shadows in the opposite direction of the main light source under the nose, under the jaw, inside the hair, and at clothing overlaps; shadows alone should not make the color muddy. Highlights are not uniformly attached to the forehead, nose, cheeks, lips, and shoulders, changing intensity according to the face direction. Do not generate extra rim lights, lens flares, rainbows, glows, or multi-directional shadows. Color/Tone: A natural color scheme centered on cream, bright beige, blue jeans, and cool gray, with suppressed saturation. Only the skin flush serves as a small warm tone. Whites do not blow out, retaining rib grooves, and blacks are not crushed, making hair bundles readable. Do not make the blue too vivid, and do not add overall yellow-green or magenta color casts. The brightness difference between the background and the skin is gentle. Highlights, midtones, and dark areas retain gradations, preventing skin whiteout, hair black-crush, or loss of clothing texture. Local contrast is restrained and placed on the face and clothing, without excessively darkening the background to make the character look like a cutout. Do not destroy the white balance of the main light source, maintaining the color relationship between the skin, white-based fabrics, and the gray background. Do not uniformly lift saturation or clarity; create the tone of a natural smartphone photo. Materials/Fine Textures: The skin retains fine pores, peach fuzz, thin red on the cheeks, and fine vertical lines on the lips. Hair is not limited to single fine lines, but coexists as hair bundle masses and scattered wisps. Ribbed knit is not uniform CG stripes; the spacing of the stretched parts changes slightly. Jeans have a fine weave and a soft feel of use; concrete has powdery fine irregularities. Do not eliminate textures with excessive sharpness or noise removal. Do not unify the smoothness of skin, hair, fabric, metal, stone, wood, plants, etc.; distinguish the reflectivity and details of each surface. Skin is not covered with uniform grain, and pore visibility changes according to highlights and shadows. Clothing weave or ribs change direction along the curves of the body and do not penetrate background straight lines. Avoid extreme JPEG compression breakdown while not being completely flawless like commercial CG, retaining the fine graininess of a natural shot. Image Processing/Completion: Completion is like a real-life smartphone photo. Do not use excessive HDR, strong skin beautification, facial contour liquefaction, eye enlargement, tooth uniformization, background synthesis feel, anime style, or illustration style. Do not over-sharpen outlines; the face and main clothing retain sufficient resolution. Suppress chromatic aberration, false colors, halos, and double outlines, retaining natural lens softness. The tone curve is rich in midtones, without clipping the black and white ends. The overall image is unified with the same light, color, texture, and graininess. To not damage the atmosphere of the scene where the woman in the cream top on the concrete stairs leans close to her body, first adjust the low-frequency large light and dark, then adjust the medium shapes of the face, clothes, and background, and finally adjust fine elements such as hair or fabric patterns. Emphasize details first without changing the overall center of gravity, and when shrinking, also maintain the face, pose, clothing color, and main background lines in the same order of recognition as the attached image. Do not locally high-define the face; give continuous resolution to the neck, shoulders, hands, and clothing. Do not strongly blur the background to make the character look like a cutout, creating a continuous focus transition according to the shooting distance. In outline processing, handle the four types separately: hair and background, skin and clothing, clothing and background, and hands and contact surfaces. The outer edge of the hair retains fine wisps, while not creating white borders or transparent gaps. The cheeks or jaw are not cut with sharp lines, but show shape with soft brightness changes along the roundness of the skin. Clothing edges are accompanied by fine shadows that convey the thickness of the fabric, preventing background lines from appearing through the body. At the overlap of hands and the lower end or clothing, do not exaggerate the presence of fingertips, only placing small drop shadows at the visible contacts. Facial identity is maintained not only by the outline but also by the combination of the distance between eyes and eyebrows, the length from the bridge of the nose to the tip, the width of the nostrils, the thickness of the upper and lower lips, the jaw distance, and the parting of the front hair. Do not rely on a single beauty template, retaining the small left-right differences of the attached image. Skin tone is continuous across the face, near the ears, neck, shoulders, and arms, without locally creating color differences like different characters. Corrections to the eyes or corners of the mouth are limited to the range that does not change emotions, without adding makeup to strengthen the impression. Adultness is expressed through a steady bone structure and natural body proportions, not deformed into a young face. Clothing reproduction is created not only by color matching but also in three stages: light-receiving parts, penumbra, and dark areas of folds. Ribs or weaves slightly change spacing and direction along human curves and poses, not as uniformly printed patterns. At points of tension change such as shoulders, armpits, chest, and waist, change the length and depth of folds. Clothing outlines are not thinned for the purpose of emphasizing body shape, maintaining the room and thickness of the attached image. When small items are present, they are not given highlights or saturation more prominent than the person, reflecting small pressure and shadows from contact with the fabric. Background perspective unifies main lines confirmed on the screen such as stairs, walls, shelves, handrails, and window frames from the camera position. Do not unnaturally bend straight lines to avoid the character's outline, or change the direction of disappearance from left to right. Background blur is not used to hide shape errors; main steps or surface directions are readable. Structures continuing outside the frame are cut off naturally in the middle, without imagining the invisible front to complete them. Dark areas of the background do not create non-existent objects or face-like patterns. In light verification, confirm whether catchlights in both eyes, bright parts of the nose, jaw shadows, hair sheen, clothing folds, and background light and dark can be explained by a single main light source. Reflected light is weaker than the main light, and shadow outlines have a softness that matches the size of the light source. Skin highlights are not painted white, retaining color and fine surface. Dark areas of hair retain the minimum gradation of discernible bundles, without painting black to eliminate noise. Bright cloth or walls also have exposure where details do not vanish. In final confirmation, compare the person's position, occupancy, background lines, and color center of gravity in full display; compare eyes, eyebrows, nose, lips, skin, and front hair in enlarged face display. Compare stitching, patterns, folds, and overlaps in clothing enlargement; compare number, contact, and shadows in hand or small item enlargement. If any part deviates from the relationship in the attached image, it will not be adopted as another aesthetic expression, and only that element will be returned to its original position, shape, color, and material. None of it is reinterpreted, and consistency is improved within the range of observable evidence. Immediately before output, divide the screen into three vertical bands: top, center, and bottom; the top confirms the top margin of the head and background vacancies, the center confirms both eyes, nose, mouth, shoulders, and clothing overlaps, and the bottom confirms hands, small items, clothing cut-off, and margins. When further divided into left and right halves, compare brightness, color, and line density with the attached image to confirm there are no traces of moving only the background to center it. Even if the center of the face is within the specified range, if the visual center of gravity including hair or shoulders is shifted, correct it. If it is correct when enlarged but the person floats in full display, adjust contact shadows, background focus, and local contrast. Even to increase naturalness, do not add moles, scars, accessories, clothing patterns, or background objects not in the image. The priority order of reproduction is character identity, pose and screen position, clothing and contact, background arrangement, light, color, fine texture; subsequent texture improvements do not change previous shapes or positions. Failure Avoidance/Prohibitions: Fix the same woman, same facial features, same hair, same clothing, same pose, same staircase arrangement, same light and color, same smartphone completion, without changing to another person or a different combination. Do not mirror the face left and right, and do not break the number of eyes, fingers, strap connections, or clothing edges. Do not add text, logos, watermarks, signatures, extra people, or mirrored characters. Do not create legs, shoes, furniture, windows, or personal belongings outside the frame. Do not change the same woman, facial features, sense of age, hairstyle, clothing, pose, contact, background arrangement, camera position, light source, color, material, or completion. Do not reverse the body or clothes left and right. Do not create extra faces, eyes, ears, arms, hands, fingers, legs, shadows, or reflections. Avoid finger fusion, joint hyperextension, poor neck connection, shoulder distortion, disappearance of clothing stitching, or background lines penetrating the body. Absolutely do not add text, numbers, subtitles, logos, brands, watermarks, signatures, or borders. Invisible elements remain unknown and are not generated.
短视频创作用法
- 做封面：保留主体、光线、构图、色彩关键词，改成自己的账号主题。
- 做AI视频：把静态画面拆成 3-5 个镜头，补充运镜、动作和转场。
- 做图文/口播：用提示词生成视觉钩子，再配合标题和脚本提高点击率。