图像转描述

分析图像并生成详细的、可用于复现的AI图像生成描述。

工作流程

步骤 1：类别检测首先，将图像分类到以下类别之一：

• 肖像— 以人物为主要主体（照片、艺术品、数字艺术）
• 风景— 自然风光、城市景观、建筑、户外环境
• 产品— 商业产品照片、商品
• 动物— 以动物为主要主体
• 插图— 图表、信息图、UI模型图、技术图纸
• 其他— 不符合以上类别的图像

步骤 2：特定类别分析根据检测到的类别生成详细的描述。

使用方法

基本分析

# Analyze an image (auto-detect category)
openclaw message send --image /path/to/image.jpg "Analyze this image and generate a detailed prompt for reproduction"

指定输出格式

自然语言（默认）

Analyze this image and write a detailed, flowing prompt description (600-1000 words for portraits, 400-600 for others).

结构化JSON

Analyze this image and output a structured JSON description with all visual elements categorized.

包含维度提取

请求维度高亮以获取每个视觉方面的标记短语：

Analyze this image with dimension extraction. Tag phrases for: backgrounds, objects, characters, styles, actions, colors, moods, lighting, compositions, themes.

类别特定元素

肖像分析涵盖：

• 模特/风格: 摄影类型、质量水平、视觉风格
• 主体: 性别、年龄、种族、肤色、体型
• 面部特征: 眼睛、嘴唇、脸型、表情
• 发型: 颜色、长度、样式、分型
• 姿势: 身体位置、朝向、腿/手的位置、视线
• 服装: 类型、颜色、图案、合身度、材质、风格
• 配饰: 珠宝、包袋、帽子等
• 环境地点、地面、背景、氛围
• 光照：类型、时间、阴影、对比度、色温
• 相机：角度、高度、镜头类型、焦距、景深、透视
• 技术：真实感、后期处理、分辨率

景观分析涵盖：

• 地形和水体特征
• 天空和大气元素
• 前景/背景构图
• 自然光照和氛围
• 调色板和摄影风格

产品分析涵盖：

• 产品特征和材质
• 设计元素和形状
• 布景和背景
• 影棚灯光设置
• 商业摄影风格

动物分析涵盖：

• 物种识别和斑纹特征
• 姿态与行为
• 表情与性格
• 栖息地与场景
• 野生动物/宠物摄影风格

插画分析涵盖：

• 图表类型（流程图、信息图、用户界面等）
• 视觉元素（图标、形状、连接线）
• 布局与层级
• 设计风格（扁平化、等轴测等）
• 配色方案与含义

输出示例

自然语言输出（肖像类）

{
  "prompt": "A stunning photorealistic portrait of a young woman in her mid-20s with fair porcelain skin and warm pink undertones. She has striking emerald green almond-shaped eyes with long dark lashes, full rose-colored lips curved in a subtle confident smile, and an oval face with high cheekbones..."
}

结构化输出（肖像类）

{
  "structured": {
    "model": "photorealistic",
    "quality": "ultra high",
    "style": "cinematic natural light photography",
    "subject": {
      "identity": "young beautiful woman",
      "gender": "female",
      "age": "mid 20s",
      "ethnicity": "European",
      "skin_tone": "fair porcelain with pink undertones",
      "body_type": "slim athletic",
      "facial_features": {
        "eyes": "emerald green, almond-shaped, intense gaze",
        "lips": "full, rose pink, subtle smile",
        "face_shape": "oval with high cheekbones",
        "expression": "confident and serene"
      },
      "hair": {
        "color": "warm honey blonde",
        "length": "long",
        "style": "soft waves",
        "part": "center"
      }
    },
    "pose": {
      "position": "standing",
      "body_orientation": "three-quarter turn to camera",
      "legs": "weight on right leg, relaxed stance",
      "hands": {
        "right_hand": "resting on hip",
        "left_hand": "hanging naturally at side"
      },
      "gaze": "direct eye contact with camera"
    },
    "clothing": {
      "type": "flowing maxi dress",
      "color": "dusty rose",
      "pattern": "solid",
      "details": "V-neckline, cinched waist, silk material",
      "style": "romantic feminine"
    },
    "accessories": ["delicate gold necklace", "small hoop earrings"],
    "environment": {
      "location": "outdoor garden",
      "ground": "cobblestone path",
      "background": "blooming roses, soft bokeh",
      "atmosphere": "dreamy and romantic"
    },
    "lighting": {
      "type": "natural sunlight",
      "time": "golden hour",
      "shadow_quality": "soft diffused shadows",
      "contrast": "medium",
      "color_temperature": "warm"
    },
    "camera": {
      "angle": "slightly below eye level",
      "camera_height": "chest height",
      "shot_type": "medium shot",
      "lens": "85mm",
      "depth_of_field": "shallow",
      "perspective": "slight compression, flattering"
    },
    "mood": "romantic, confident, ethereal",
    "realism": "highly photorealistic",
    "post_processing": "soft color grading, subtle glow",
    "resolution": "8k"
  }
}

含尺寸信息

{
  "prompt": "...",
  "dimensions": {
    "backgrounds": ["outdoor garden", "blooming roses", "soft bokeh"],
    "objects": ["delicate gold necklace", "small hoop earrings"],
    "characters": ["young beautiful woman", "mid 20s", "European"],
    "styles": ["photorealistic", "cinematic natural light photography"],
    "actions": ["standing", "three-quarter turn", "direct eye contact"],
    "colors": ["dusty rose", "honey blonde", "emerald green"],
    "moods": ["romantic", "confident", "ethereal", "dreamy"],
    "lighting": ["golden hour", "natural sunlight", "soft diffused shadows"],
    "compositions": ["medium shot", "85mm", "shallow depth of field"],
    "themes": ["romantic feminine", "portrait photography"]
  }
}

最佳效果建议

1. 高分辨率图像能生成更详细的提示词
2. 清晰、光照良好的图像能实现更好的类别检测
3. 当您需要对单个元素进行程序化访问时请求结构化输出
4. 使用尺寸提取功能构建提示数据库或训练数据时
5. 若需要，请明确自然语言输出的字数要求以便进行

集成

此技能适用于任何具备视觉功能的模型。为达到最佳效果，推荐使用：

• GPT-4视觉版
• Claude 3（Opus/Sonnet版）
• Gemini Pro视觉版