Prompts.SkinAI Image Prompts

粉绿双色梦核时尚大片风格化

ModelGPT-Image2
ParametersBeeldverhouding 1:1
Prompt

Please perform stylized creation based on the image I uploaded, instead of simple filter processing. Retain the identity characteristics, core posture, main contours, and pictorial relationships of the original subject, but allow for the redesign of scenes, lighting, colors, space, and visual narrative. Transform the scene into a pink and green dual-color dreamcore fashion blockbuster: the overall look is like a mixture of 90s retro fashion photography, Y2K experimental video, indie music album covers, and surreal advertising visuals. Use high-recognition contrasting pink and turquoise colors: pink sky, rosy skin highlights, peach pink air haze, turquoise shadows, lake green distant views, and mint green environmental reflections. The overall colors are soft yet impactful, like faded film superimposed with dreamlike imagery. Please reconstruct the scene environment: you can place the original subject in a more open, more surreal space, such as a pink desert, turquoise mountains, a dreamlike studio, a minimalist room, a light-colored pool, beside a retro sports car, a cloud horizon, a soft mist glass space, or a futuristic street scene. Visual language: soft focus, film grain, low contrast, soft mist light, slight overexposure, retro magazine print feel, slightly distorted colors, dreamlike spatial layers. The subject is clear, and the environment is like a memory, a dream, or an advertising set. Overall, it should not look like a random AI image, but like a complete art direction work: unified color palette, prominent subject, clean composition, negative space, a sense of high-end quality, and a publishable visual hook. Prohibited: Do not directly copy the 3x3 grid of the reference image; do not generate multiple small images; do not add meaningless text; do not show watermarks; do not generate real brand logos; do not let characters' facial features collapse; do not let the subject disappear; do not just apply a pink filter.