Kiss-face character portrait template: GPT Image 2 prompt

Designated character: {Frieren}
Character source / worldview: {optional, e.g. Genshin Impact / Honkai: Star Rail / original character / a self-made worldview}
Generate a 9:16 vertical, ultra-realistic close-up phone portrait. The main subject is a real-life version of {designated character}, a clearly adult female character who does not resemble any real celebrity, influencer, or famous person.
Preserve the most core recognizable features of {designated character}: hairstyle, hair color, eye color, signature color, iconic accessories, clothing elements, temperament, expression, and worldview symbols, but translate the whole into a believable, high-end real-life character portrait look, without a cheap cosplay feel.
The shot is an ultra-close selfie / friend's-phone close-up perspective, around 24mm phone wide-angle, lens very close to the face, the face filling most of the frame. Composition runs from the top of the head to around the chest; the subject faces the camera directly, as if she suddenly leaned in to give the viewer a cute pursed-lips kiss.
Expression focus: a clear pursed-lips kiss face, lips gently pursed forward into a round natural kissing shape; lips moist with highlights and fine texture. Eyes looking at the camera, slightly squinted, with a playful, coquettish, intimate, cute teasing feeling. Cheeks softly flushed, expression natural, not exaggerated, no cartoonish faces. Use the kiss only as an expression guide, do not generate a floating lipstick-mark sticker or emoji graphic.
Refined translucent makeup: creamy soft-glow skin with real texture preserved; small highlights on the nose bridge, cheeks, and lip peaks; pink dewy lips with a prominent lip bead and clear reflection; clean eye makeup, thin eyeliner, distinct lashes, faint pearl shimmer at inner corners and lower lash line. Overall a real-photo portrait quality, no plastic skin.
Hairstyle auto-translated from {designated character}, preserving hair color, bangs, ornaments, highlights, or contour features, but with realistic material, clear strands, natural fluffiness and fine flyaways. A few strands fall naturally beside the face or in front of the shoulder to enhance close-up intimacy.
Clothing auto-designed into realistically wearable high-end character-portrait attire based on the character's signature color and identity: a futuristic shirt, functional harness, short jacket, refined uniform-style top, stage-style casual wear, or cyber-idol look. Clothing complete, clean, layered, keeping character recognizability, but not copying cheap cosplay outfits.
Scene auto-matched to the character's worldview: a cyber studio, game room, backstage dressing room, futuristic rest pod, nighttime room, urban neon space, performance backstage, or a modern private space in the character's world. Background blurred, but keep a few recognizable lights, screens, devices, walls, props, or symbols so the character isn't floating in empty space.
Lighting: subject lit by soft frontal fill, clear highlights on face, eyes, nose tip, and lips; background low-exposure and blurred, ambient light from the side-rear outlining hair strands, shoulders, and face edges. Subject noticeably brighter than the environment, creating a close-up glow without overexposing skin detail.
Image quality: ultra-realistic phone close-up, slight SNS compression, soft-focus lens, shallow depth of field, background bokeh, clean warm-cool mixed light, clear skin and lip detail, clear hair strands, overall cute, intimate and real, like the character leaning close to give the viewer a kiss.
Naturally add a low-key small handwritten signature 'Zhong' in the lower-right corner as the designer's mark, not stealing focus.
Negative prompt: no underage feel, no infantile face, no explicit pornography, no isolated body-part close-ups, no exaggerated anime face, no lip deformity, no exposed teeth, no lipstick-mark sticker over the face, no floating emoji, no low resolution, no plastic skin, no over-smoothing, no cheap cosplay, no misaligned features, no misaligned eyes, no garbled text, no big logo, no multiple people, no collage.