OpenZhenZhen AI 提示词画廊 · gpt-image-2.5 · image

复古一次性相机卧室抓拍自拍

Retro Disposable Camera Bedroom Candid Selfie

A close candid bedroom selfie of the same adult subject from the supplied reference, rendered as a late-1990s disposable-camera snapshot with direct flash, strong grain, motion ghosting, crooked framing, and a relaxed sleepy expression.

复古一次性相机卧室抓拍自拍 — AI image creation
复古一次性相机卧室抓拍自拍

提示词 / Prompt

中文提示词

将上传参考图中的同一成年主体转换为一张复古一次性相机卧室抓拍自拍。主体随意躺在未整理的床上，头靠枕头，面向镜头露出放松而困倦的微笑，头发因刚刚的动作略微飞动并被抓拍；镜头距离极近，脸部约占画面一半。主体穿完整、舒适且不暴露的条纹缎面成人家居服，配色为淡紫色、粉色和浅蓝色，佩戴精致银色项链和小耳钉，以 ref-10a-v1 保持人物身份线索一致。模拟真实的1990年代后期一次性胶片相机：生硬的内置直闪、脸部强烈热点、自然皮肤油光、明显闪光衰减、昏暗偏泥的背景、可见35mm颗粒、廉价塑料镜头柔化、边缘轻微拖影、动作造成的运动模糊、手部轻微重影、不均匀曝光和发白的黑位。背景为舒适卧室，褶皱白床单、浅色枕头、床头柜、昏暗台灯、少量散落个人物品与半拉窗帘。构图歪斜、轻微荷兰式倾斜和对角线取景，主体略偏离中心，故意保留业余抓拍与不完美裁切。加入尘埃颗粒、淡淡扫描线、褪色照片纹理、划痕、轻微漏光和棱镜光晕，但仍保持面部可辨认。4:5竖幅。负面约束：假人姿势、完美模特凝视、僵硬微笑、僵硬手势、扁平面部、蜡质皮肤、无法辨认的脸、暴露或滑落的服装、过度摆拍、电影感单反景深、水印、文字和签名。

English prompt

Transform the same adult subject from the supplied reference into a retro disposable-camera bedroom snapshot selfie. The subject lies casually on an unmade bed with their head against a pillow, looking toward the camera with a relaxed sleepy smile; their hair is slightly caught in motion from a recent movement. Shoot extremely close to the lens, with the face occupying about half the frame. The subject wears complete, comfortable, non-revealing striped satin adult sleepwear in lavender, pink, and baby blue, with a delicate silver necklace and small stud earrings; use ref-10a-v1 to keep the subject's identity cues consistent. Simulate a real late-1990s disposable film camera: harsh built-in direct flash, a strong facial hotspot, natural skin shine, pronounced flash falloff, a dark muddy background, visible 35mm grain, cheap plastic-lens softness, slight edge smearing, accidental motion blur, subtle hand ghosting, uneven exposure, and lifted blacks. Set the scene in a cozy bedroom with rumpled white sheets, pale pillows, a bedside table, a dim lamp glow, a few scattered personal items, and partly drawn curtains. Use crooked framing, a slight Dutch tilt and diagonal composition, with the subject slightly off-center; preserve an amateur candid feeling and intentionally imperfect crop. Add dust particles, faint scan lines, faded print texture, scratches, slight light leaks, and a restrained prism flare while keeping the face recognizable. Vertical 4:5 frame. Negative constraints: mannequin pose, perfect model stare, stiff smile, stiff hands, flat face, waxy skin, unrecognizable face, exposed or slipping clothing, overly staged posing, cinematic DSLR depth of field, watermark, text, and signature.

查看 API 模型

探索更多图片和视频提示词 · 生成图片 · 生成视频 · 在 AI 画布中创作