返回 Prompt 库



南亚女郎背影回眸写真
{ "prompt": "Using the uploaded image as exact face and body reference with ultra face consistency and identity preservation, create an ultra-realistic, vertical 2:3 candid cinemat




Prompt
{
"prompt": "Using the uploaded image as exact face and body reference with ultra face consistency and identity preservation, create an ultra-realistic, vertical 2:3 candid cinematic fashion portrait photography image of a glamorous South Asian woman viewed from a graceful back-side angle. She is standing in an elegant, poised posture with her back mostly facing the camera while turning her head softly over her shoulder toward the viewer with a serene, confident expression. Her long black wavy hair cascades naturally down her back.\n\nShe is wearing a luxurious modern couture saree in shimmering metallic champagne-light gold fabric with rich reflective textures. The saree features extensive heavy gold embroidery with intricate swirling botanical and floral patterns throughout the drape and lower pleats. The blouse is a jewellery-integrated couture bustier fully encrusted with clustered diamonds and large emerald gemstones. Delicate multi-layered jeweled chains drape elegantly across her bare shoulder blades and upper arms.\n\nShe is heavily accessorized with layered diamond and emerald necklaces, oversized cascading earrings, a statement emerald ring, and additional fine luxury jewelry. In her left hand she holds a whimsical luxury clutch designed like a golden jeweled net bag containing a single ripe mango.\n\nLighting is dramatic cinematic editorial lighting with warm highlights emphasizing the sparkle of the jewels, metallic sheen of the saree fabric, and contours of the body. The background is dark, moody, softly blurred, and luxurious with textured shadowy walls and an out-of-focus arrangement of pale pink and white flowers placed subtly behind her. Depth of field is shallow with creamy bokeh. High-end Vogue-style fashion photography aesthetic, photorealistic skin texture, premium couture styling, rich contrast, cinematic shadows, ultra detailed, 8k quality.\n\nBody proportions are glamorous and curvy with a fuller bust silhouette while maintaining elegance and realism.\n\nAdd elegant cursive signature text 'XSydneyFan' in the bottom corner.",
"negative_prompt": "low quality, blurry, distorted anatomy, extra limbs, duplicate body parts, bad hands, asymmetrical eyes, cartoon, anime, overexposed lighting, flat lighting, poorly drawn jewelry, messy saree draping, watermark, text artifacts, low resolution, unrealistic skin, stiff pose, cropped face, deformed fingers",
"aspect_ratio": "2:3",
"style": "ultra-realistic cinematic fashion editorial",
"camera": {
"angle": "back-side over-shoulder portrait angle",
"lens": "85mm portrait lens",
"depth_of_field": "shallow DOF with creamy bokeh",
"quality": "8k photorealistic"
},
"lighting": {
"type": "cinematic studio lighting",
"mood": "dark luxurious moody ambiance",
"highlights": "warm highlights on jewelry and metallic fabric"
}
}
如何使用这个 Prompt
- 用上方的复制按钮复制完整提示词。
- 把方括号 [ ] 里的占位内容替换成你自己的主体、场景或细节。
- 粘贴到 Gemini / Nano Banana Pro 生成,或直接点上面的「试试这个 Prompt」在本站免费生成,再对照示例图微调用词。
风格说明
- 3D Render
- CGI 质感:表面平滑、全局光照可控,纵深像 Blender/C4D 渲染而非照片。
- Photo
- 要求模型输出照片级真实感:自然光线、可信的皮肤与材质、类似相机的景深,而不是插画或 3D 质感。
- Typography
- 文字本身是画面的一部分。Nano Banana Pro 这类新模型能渲染清晰的美术字,但提示词里要把具体文字写短、并用引号标出。
- Cinematic
- 电影感渲染:宽画幅构图、强烈的光比、调色(常见青橙对比),画面自带叙事氛围。
- Anime
- 日式动画审美:线条干净、眼睛大而传神、赛璐璐上色,场景语言来自动画而非摄影。
- Cartoon
- 造型简化、比例和表情夸张、平涂或赛璐璐上色——画面易读,角色风格也容易保持一致。
- Hero Portrait
- 把单一主体拍成主视觉:紧凑构图、有力的眼神或姿态,背景简化,让人物撑起整张画面。
- Fashion Look
- 时尚大片式造型:突出服装细节与面料质感,模特姿态和布光借鉴杂志拍摄。
- Vehicle
- 汽车渲染:漆面反光、轮毂与格栅细节,借鉴汽车广告的低机位戏剧性视角。
- Nature Landscape
- 广阔的自然风光构图:层次纵深、天气与时间(晨昏)描述承担大部分氛围表达。
常见问题
这个 Prompt 怎么在 Gemini(Nano Banana Pro)里使用?
点击上方的复制按钮拿到完整提示词,粘贴到 Gemini 或任何支持 Nano Banana Pro 的工具里,再对照本页示例图检查效果。也可以直接点「试试这个 Prompt」在本站生成——每天有免费次数,无需注册。
可以修改「南亚女郎背影回眸写真」这个提示词吗?
可以。方括号 [ ] 里的内容是占位参数,本来就应该替换成你自己的主体或细节;主体、光线、风格关键词也都可以自由调整,措辞上的小改动往往带来明显不同的结果。
这个提示词适用于哪些 AI 模型?
它最初是为 Google 的 Gemini 图像模型(含 Nano Banana Pro)分享的。大多数文生图模型也能理解,但构图和文字渲染质量会因模型而异。





