Nano Banana 图像生成提示词架构（JSON 格式）

# Nano Banana 图像生成提示词规则

使用 Nano Banana 生成图像时，提示词必须**始终**严格遵循以下定义的 JSON 格式。

## 提示词 Schema

```json
{
  "image_type": "定义源图像的特定媒介、艺术风格或格式。",
  "time_period_and_year": "根据照片中可见的时尚、技术、图像质量和色彩分级，估算具体的年份或十年。",
  "mood_and_vibe": "描述图像唤起的情感氛围、能量和无形的“感觉”（例如，它给人的特定心理印象）。",
  "subject": "描述主要人物，重点关注人口统计学特征、身体形态、独特的身体特征和姿势。",
  "clothing": "详细描述服装，严格指定服装名称、面料纹理、图案、颜色以及衣服在主体上的合身程度。",
  "hair": "描述头发颜色、具体的发型名称、长度和纹理。",
  "face": "描述面部特征、皮肤纹理、妆容细节和确切的面部表情。",
  "accessories": "列出所有可见的配饰、珠宝、眼镜或手持物品，包括它们的材质和设计细节。",
  "action": "描述场景中发生的具体活动、动作或互动。",
  "location": "描述环境、可见的背景元素、建筑风格和空间背景。",
  "lighting": "分析光源、方向、色温、光的硬度/柔和度以及阴影特征。",
  "camera_angle_and_framing": "描述相机相对于主体的垂直和水平角度（例如，视线水平、低角度）以及拍摄构图大小。",
  "camera_equipment": "估算可能的相机类型、镜头焦距（例如，广角 vs 长焦）、光圈效果（景深）以及胶片类型或数字传感器特性。",
  "style": "描述整体美学、调色板、艺术技巧和视觉处理风格。",
  "negative_prompt": "列出要排除的视觉缺陷或不需要的元素，以确保高质量。"
}
```

## 说明

1.  **输出格式**：输出必须是单个有效的 JSON 对象。
2.  **完整性**：所有字段都是必需的。如果某个字段不适用，请提供一个合理的默认值或将其描述为“中性”或“标准”。
3.  **细节**：在每个字段中都应详细且具体，以确保生成高质量的图像。
4.  **语言**：JSON 中的值应为英文（因为大多数图像生成模型都针对英文提示词进行了优化），除非另有说明。

---

**Original English:**
# Nano Banana Image Generation Prompt Rules

When generating images using Nano Banana, the prompt must ALWAYS strictly follow the JSON format defined below.

## Prompt Schema

```json
{
  "image_type": "Define the specific medium, art style, or format of the source image.",
  "time_period_and_year": "Estimate the specific year or decade based on fashion, technology, image quality, and color grading visible in the photo.",
  "mood_and_vibe": "Describe the emotional atmosphere, energy, and intangible 'feeling' evoked by the image (e.g., the specific psychological impression it gives).",
  "subject": "Describe the main character(s) focusing on demographics, body morphology, distinct physical features, and posture.",
  "clothing": "Describe the outfit in detail, strictly specifying garment names, fabric textures, patterns, colors, and how the clothes fit on the subject.",
  "hair": "Describe the hair color, specific hairstyle name, length, and texture.",
  "face": "Describe facial features, skin texture, makeup details, and the exact facial expression.",
  "accessories": "List all visible accessories, jewelry, glasses, or held items, including their material and design details.",
  "action": "Describe the specific activity, movement, or interaction occurring in the scene.",
  "location": "Describe the environment, visible background elements, architectural style, and spatial context.",
  "lighting": "Analyze the light source, direction, color temperature, hardness/softness of the light, and shadow characteristics.",
  "camera_angle_and_framing": "Describe the vertical and horizontal angle of the camera relative to the subject (e.g., eye-level, low-angle), and the shot composition size.",
  "camera_equipment": "Estimate the likely camera type, lens focal length (e.g., wide-angle vs telephoto), aperture effect (depth of field), and film stock or digital sensor characteristics.",
  "style": "Describe the overall aesthetic, color palette, artistic technique, and visual processing style.",
  "negative_prompt": "List visual defects or unwanted elements to be excluded to ensure high quality."
}
```

## Instructions

1. **Output Format**: The output must be a single valid JSON object.
2. **Completeness**: All fields are required. If a field is not applicable, provide a reasonable default or describe it as "neutral" or "standard".
3. **Detail**: Be descriptive and specific in each field to ensure high-quality image generation.
4. **Language**: The values in the JSON should be in English (as most image generation models are optimized for English prompts), unless otherwise specified.