韩系冷蓝夜街CCD直闪人像写真
韩系冷蓝夜街CCD直闪人像写真，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
9:16, CCD/point-and-shoot camera direct flash, Korean night street solo portrait, cold blue low-light subculture, on-site snapshot between a vending machine and a fence. Adult Korean female, cold white to ivory skin tone, clean oval face, elongated upturned eyeliner, coral pink pouty lips, real skin texture, natural semi-matte base makeup, black medium-long straight hair slightly messy, a few strands across the side of the face. Wearing a black long-pile fuzzy coat, teal thin satin camisole underneath with a black lace neckline, black pants and belt, bright silver thin necklace and small earrings. The person is located next to a cold blue vending machine, with colorful canned drinks arranged inside the vending machine on the back left, and a chain-link fence and deep indigo night street on the back right. Close-up eye-level half-body to three-quarter body composition, body slightly turned to one side of the frame, right hand holding a silver-black canned drink and raising it near the cheek and chin, the other hand hidden in the fuzzy coat, head slightly tilted, eyes half-drooping looking sideways and downwards, a cold, blurry, and slightly weary expression. On-axis hard direct flash directly brightens the subject, clear highlights formed on the catchlight, bridge of the nose, hair strands, can body, satin camisole, and the edges of the coat, controlled highlights, cheeks maintain soft diffuse reflection, the vending machine provides cold blue local ambient light, background rapidly decays and darkens, slight CCD noise and low-light night shot feel. Add artistic white handwritten text 'ChenLin' as a signature in the bottom right corner of the frame. No plastic skin, continuous oil film across the whole face, excessive skin smoothing, deformed fingers, disconnected head and neck, extra figures, background stealing the spotlight, extra text watermarks.
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。