📝 英文生图提示词
{ "prompt": "使用上传的参考图片,**100% 保留人物的面部特征、身份、年龄、肤色和表情**。应用**3D 动漫水彩风格**,同时保持面部特征不变。人物拥有波浪形深棕色长发(或根据参考图调整)、白皙皮肤、大而富有表现力的眼睛、红润的脸颊、俏皮或温柔的表情。服装:宽松休闲上衣({argument name=\"top color\" default=\"粉色 T 恤\"} 或参考图启发),高腰裤,鞋子可选。姿势:自然坐姿或站姿,一只手拿着一个物体({argument name=\"held object\" default=\"棒棒糖、饮料或个人物品\"}),另一只手放松。背景:户外都市或柔和的抽象水彩环境,柔和的大气光线,暖色调或柔和的粉彩色调,带有微妙散景的绘画纹理。高度细节,电影感,鲜艳色彩,柔和景深,杰作级品质。", "negative_prompt": "模糊,低质量,变形,多余肢体,糟糕的解剖结构,水印,文字叠加,写实照片,3D 渲染,卡通化(非动漫),过曝,欠曝,丑陋的脸,糟糕的手,面部改变,身份改变", "parameters": { "steps": 50, "cfg_scale": 7.5, "sampler": "Euler a or DPM++ 2M Karras", "strength": 0.55, "noise": 0.3, "width": 832, "height": 1216, "aspect_ratio": "2:3" }, "usage_instructions": "在 img2img 模式下上传您的参考图片。AI 将**完全保留参考图中的面部特征**,同时应用动漫 3D 水彩风格、服装和环境。将强度保持在 0.5-0.6 左右,以实现完美的脸部匹配和风格化。略微增加强度可实现轻微的姿势或背景变化。" }
{ "prompt": "Use the uploaded reference image to **preserve 100% of the facial features, identity, age, skin tone, and expression** of the person. Apply a **3D anime watercolor style** while keeping the face identical. The person has long wavy dark brown hair (or adapted from reference), fair skin, big expressive eyes, flushed cheeks, playful or gentle expression. Clothing: loose casual top (pink t-shirt or reference-inspired), high-waisted pants, shoes optional. Pose: sitting or standing naturally, one hand holding an object (lollipop, drink, or personal prop), other hand relaxed. Background: outdoor urban or soft abstract watercolor environment, soft atmospheric lighting, warm or pastel tones, painterly texture with subtle bokeh. Highly detailed, cinematic, vibrant colors, soft depth of field, masterpiece quality.", "negative_prompt": "blurry, low quality, deformed, extra limbs, bad anatomy, watermark, text overlay, realistic photo, 3d render, cartoonish (non-anime), overexposed, underexposed, ugly face, bad hands, face altered, identity changed", "parameters": { "steps": 50, "cfg_scale": 7.5, "sampler": "Euler a or DPM++ 2M Karras", "strength": 0.55, "noise": 0.3, "width": 832, "height": 1216, "aspect_ratio": "2:3" }, "usage_instructions": "Upload your reference image in img2img mode. The AI will **keep the face exactly as in the reference**, while applying the anime 3D watercolor style, clothing, and environment. Keep strength around 0.5-0.6 for perfect face match with stylization. Increase slightly for minor pose or background variations." }💡 提示词说明:本案例采集自 OpenNana 创意库,精选高清视觉灵感,适用 ChatGPT / Midjourney 模型。
🔮 画面意境解析:完美复刻真人的3D水彩动漫画风。
