ComfyUI Globetrotter Nodes

A comprehensive collection of custom ComfyUI nodes for generating culturally diverse AI image prompts. This project features a fully data-driven architecture with gender-aware attire filtering, comprehensive body part coverage, and intelligent prompt generation optimized for AI image models.

✨ Key Features

🌍 Cultural Diversity & Authenticity

• Multi-Country Support: Dynamic nodes for countries with extensive cultural data
• Regional Variations: Detailed appearance options for different regions within countries
• Cultural Context: Activities, festivals, landmarks, and traditional colors
• Authentic Attire: Culturally accurate clothing options with detailed descriptions

👕 Advanced Attire System

• Gender-Aware Filtering: Clothing options automatically filter based on selected gender
• Comprehensive Body Coverage: Support for 18+ body parts including:
  • Arms: Forearm, Hands, Palms, Upper_Arm, Wrists
  • Head: Head, Chin, Ears, Forehead, Hair, Nose
  • Legs: Legs, Ankles, Feet
  • Upper_Body: Upper_Body, Chest, Shoulders
  • Waist: Waist area clothing
• Rich Descriptions: Each attire item includes detailed descriptions for AI prompt generation
• Smart Randomization: Intelligent random selection with contextual awareness

🤖 AI-Optimized Prompt Generation

• Multiple Prompt Styles: AI-friendly, creative, artistic, technical, and balanced modes
• Smart Defaults: Intelligent parameter combinations that work well together
• Contextual Suggestions: Weather-activity and emotion-setting pairings
• Technical Enhancements: Quality keywords, composition rules, and lighting optimization
• LLM Integration: Optional prompt rewriting with local language models

🎛️ Intuitive Controls

• Simplified UI: Clear, logical control grouping with helpful tooltips
• Universal Options: Every dropdown includes "none" and "random" options
• Smart Randomization: Four levels of randomization (off, light, moderate, full)
• Reproducible Results: Seed-based consistency for repeatable outputs
• Gender-Aware Interface: Attire options automatically filtered by gender selection

📊 Data-Driven Architecture

• Fully Modular: All data stored in organized JSON files
• Zero Hardcoding: No attire, cultural, or prompt data in Python code
• Easy Extension: Add new countries, body parts, or attire by updating JSON files
• Robust Loading: Error-tolerant JSON loading with graceful fallbacks

📁 Project Structure

comfyui-globetrotter/
├── data/                          # Organized data files
│   ├── attire/                    # Clothing and accessories by body part
│   │   ├── arms/                  # Forearm, hands, palms, upper_arm, wrists
│   │   ├── head/                  # Head, chin, ears, forehead, hair, nose
│   │   ├── legs/                  # Legs, ankles, feet
│   │   ├── upper_body/            # Upper_body, chest, shoulders
│   │   └── waist/                 # Waist area clothing
│   ├── appearance/                # Regional appearance data
│   ├── poses/                     # Country-specific poses
│   ├── cultural/                  # Cultural activities, festivals, landmarks
│   ├── config/                    # System configuration
│   │   ├── gender_config.json     # Gender mapping and filtering rules
│   │   └── prompt_config.json     # Prompt generation settings
│   ├── generation/                # AI prompt optimization
│   │   ├── smart_defaults.json    # Intelligent default combinations
│   │   ├── contextual_suggestions.json  # Context-aware suggestions
│   │   └── ai_prompt_structure.json     # AI-optimized prompt structure
│   ├── ui/                        # User interface data
│   │   ├── dynamic_node_options.json    # UI dropdown options
│   │   ├── weather_moods.json           # Atmospheric conditions
│   │   └── complementary_colors.json    # Color palette suggestions
│   ├── styles/                    # Artistic and photographic styles
│   └── prompts/                   # LLM prompt templates
├── globetrotter_nodes/            # Core Python modules
│   ├── dynamic_nodes.py           # Main dynamic node generation
│   ├── ollama_llm_node.py         # LLM integration node
│   └── text_combiner_node.py      # Text utility node
└── requirements.txt               # Python dependencies

🚀 Installation

1. Clone the repository into your ComfyUI custom_nodes directory:

   cd /path/to/ComfyUI/custom_nodes
   git clone <repository-url> comfyui-globetrotter
   

2. Install dependencies (optional but recommended for full features):

   cd comfyui-globetrotter
   pip install -r requirements.txt
   

3. Restart ComfyUI to load the new nodes.

💡 Usage

Dynamic Country Nodes

Each country automatically gets its own node (e.g., "India Attire") with comprehensive options:

Core Parameters

• Age: Young adult, Adult
• Gender: Female, Male (with automatic attire filtering)
• Region: Country-specific regions (e.g., Java, Punjab, Rajasthan)
• Hair Style: Generic options + region-specific defaults
• Emotion: Confident, Serene, Joyful, Contemplative, etc.

Attire Selection (Gender-Filtered)

Individual dropdowns for each body part with culturally appropriate options:

• Arms: Forearm decorations, hand accessories, wrist jewelry
• Head: Headwear, face decorations, ear accessories
• Upper Body: Traditional tops, chest accessories, shoulder pieces
• Lower Body: Traditional bottoms, leg wear, foot attire
• Waist: Belts, sashes, waist decorations

Cultural Context

• Poses: Traditional and cultural poses
• Cultural Elements: Festivals, activities, traditions
• Settings: Markets, temples, landmarks, urban areas
• Atmosphere: Weather, lighting, mood combinations

Advanced Controls

• Detail Level: Basic, Detailed, Cinematic, Artistic, Photorealistic
• Composition: Rule of thirds, Centered, Close-up, Wide shot, etc.
• Prompt Optimization: AI-friendly, Creative, Technical, Artistic, Balanced
• Randomization: Off, Light, Moderate, Full (with smart contextual choices)
• Custom Elements: LoRA triggers, custom prompts
• Experimental: AI-powered prompt rewriting (requires transformers library)

Example Attire JSON Structure

data/attire/upper_body/in.json

{
  "country": "IN",
  "body_part": "upper_body",
  "attires": [
    {
      "name": "Saree Blouse",
      "type": "clothing",
      "description": "A fitted upper garment worn under a saree, often short-sleeved or sleeveless, and tailored to match the saree.",
      "material": ["cotton", "silk", "synthetic"],
      "region": ["Nationwide"],
      "gender": ["female"],
      "occasion": ["daily wear", "wedding", "festival"]
    },
    {
      "name": "Kurta",
      "type": "clothing",
      "description": "A loose-fitting, long tunic worn by both men and women, often paired with churidar or jeans.",
      "material": ["cotton", "silk", "linen"],
      "region": ["Nationwide"],
      "gender": ["male", "female", "unisex"],
      "occasion": ["daily wear", "casual", "formal"]
    }
  ]
}

data/appearance/in.json

{
  "country": "IN",
  "regions": [
    {
      "name": "Punjab",
      "description": "People from Punjab. Features include wheat-colored to medium brown skin and strong facial structure.",
      "skin_tone": "wheat-colored to medium brown",
      "hair": "black, thick and wavy"
    },
    {
      "name": "South India",
      "description": "People from Tamil Nadu, Kerala, Karnataka, and Andhra Pradesh. Features include dark to very dark skin.",
      "skin_tone": "dark to very dark brown",
      "hair": "black, thick and curly"
    }
  ]
}

Gender-Aware Attire Filtering

The system automatically filters attire options based on the selected gender:

• Female: Shows items with gender: ["female"] or gender: ["unisex"]
• Male: Shows items with gender: ["male"] or gender: ["unisex"]
• Validation: Inappropriate combinations are automatically skipped during prompt generation

Utility Nodes

Text Combiner Node

• Combines multiple text inputs into a single formatted string
• Useful for complex prompt construction workflows

Ollama LLM Node

• Advanced prompt enhancement using local Ollama language models
• Includes artistic styles, camera settings, lighting options
• Dynamic loading of style configurations from JSON files

Smart Prompt Generation

AI-Optimized Output Examples

AI-Friendly Mode:

Highly detailed, best quality, rule of thirds composition, A young adult female from Punjab, India, with a confident expression, wearing Saree Blouse: A fitted upper garment worn under a saree, often short-sleeved or sleeveless, and Churidar: Traditional fitted trousers, atmosphere: golden hour

Creative Mode:

Rule of thirds composition, Highly detailed, A young adult female from the Punjab region of India, wearing Saree Blouse: A fitted upper garment worn under a saree and Churidar: Traditional fitted trousers, with a confident expression, color palette: warm earth tones, golden natural colors, atmosphere: golden hour

🔧 Adding New Content

Adding a New Country

1. Create country entry in data/countries.json:

   {
     "name": "Country Name",
     "code": "cc",
     "flag": "🇨🇨"
   }
   

2. Add appearance data in data/appearance/cc.json:

   {
     "country": "cc",
     "regions": [
       {
         "name": "Region Name",
         "description": "Physical description",
         "skin_tone": "skin tone description",
         "hair": "hair description"
       }
     ]
   }
   

3. Create attire files in appropriate body part directories:

   • data/attire/head/cc.json
   • data/attire/upper_body/cc.json
   • data/attire/legs/cc.json
   • etc.

4. Add cultural context (optional):

   • data/poses/cc.json
   • data/cultural/cc.json

5. Restart ComfyUI to load the new country node.

Adding New Attire Items

1. Edit the appropriate JSON file (e.g., data/attire/head/in.json)
2. Add new attire object:

   {
     "name": "Attire Name",
     "type": "clothing",
     "description": "Detailed description for AI prompts",
     "material": ["cotton", "silk"],
     "region": ["Region1", "Region2"],
     "gender": ["male", "female", "unisex"],
     "occasion": ["daily wear", "formal", "festival"]
   }
   

3. Restart ComfyUI to load the new options.

Extending Body Part Coverage

1. Create new directory under data/attire/ (e.g., accessories/)
2. Add country-specific JSON files with the new body_part field
3. System automatically detects and includes new body parts in UI

⚡ Advanced Features

Experimental LLM Rewriting

• Toggle Option: Each node includes experimental_llm_rewrite
• Local Models: Uses Hugging Face distilgpt2 model when available
• Smart Filtering: Automatic repetition detection and removal
• Length Control: Output length limits based on prompt optimization mode
• Graceful Fallback: Silently skips if dependencies are missing

Intelligent Randomization

• Smart Random Mode: Context-aware random selections
• Seed Control: Fixed, increment, decrement, or system random
• Contextual Pairing: Weather-activity and emotion-setting combinations
• Optimal Defaults: Age-gender combinations that work well together

Prompt Optimization Modes

| Mode | Purpose | Style | |------|---------|-------| | AI-Friendly | Stable Diffusion, FLUX | Quality keywords first, clear structure | | Creative | Artistic generation | Narrative flow, artistic language | | Technical | Professional workflows | Precise technical terms | | Artistic | Fine art creation | Museum-quality descriptions | | Balanced | General purpose | Mix of technical and creative |

Configuration System

data/config/gender_config.json

{
  "gender_mappings": {
    "male": ["male", "unisex"],
    "female": ["female", "unisex"],
    "non-binary": ["unisex"]
  }
}

data/config/prompt_config.json

{
  "technical_enhancements": {
    "detailed": "8k resolution, highly detailed",
    "cinematic": "professional photography, cinematic lighting"
  },
  "detail_prefixes": {
    "ai_friendly": {
      "detailed": "highly detailed, best quality, "
    }
  }
}

🛠️ Technical Details

Dynamic Node Generation

• Factory Pattern: Nodes created programmatically for each country
• Closure-Based: Input types capture country-specific data
• Memory Efficient: Data loaded once and cached
• Error Tolerant: Graceful handling of missing files

Gender Validation System

• Runtime Filtering: Attire validated during prompt generation
• File Discovery: Automatic detection of correct attire file paths
• Cross-Reference: Body part mapping across directory structure
• Fallback Options: Safe defaults when validation fails

Data Loading Architecture

• Lazy Loading: JSON files loaded only when needed
• Caching: Frequent data cached in memory
• Error Recovery: Default values for missing files
• Validation: Schema checking for critical fields

📋 Requirements

Core Dependencies

torch>=1.9.0              # PyTorch for tensor operations
torchvision>=0.10.0        # Computer vision utilities
transformers>=4.0.0        # Hugging Face transformers (for LLM features)
requests>=2.25.0           # HTTP requests for Ollama API

Optional Dependencies

accelerate                 # Faster model loading
safetensors               # Secure tensor serialization

Installation:

pip install -r requirements.txt

System Requirements

• Python: 3.8 or higher
• ComfyUI: Latest version recommended
• Memory: 4GB+ RAM for LLM features
• Storage: ~50MB for full dataset

🤝 Contributing

Guidelines

1. Follow JSON Schema: Maintain consistent data structure
2. Cultural Sensitivity: Ensure authentic and respectful representation
3. Test Additions: Verify new content works across gender combinations
4. Documentation: Update relevant documentation for new features

Code Style

• Python: Follow PEP 8 conventions
• JSON: Use 2-space indentation
• Comments: Document complex logic and cultural context

Pull Request Process

1. Fork the repository
2. Create feature branch (git checkout -b feature/new-country)
3. Add your changes with appropriate tests
4. Update documentation
5. Submit pull request with detailed description

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Cultural Consultants: For authentic attire and cultural information
• ComfyUI Community: For feedback and feature requests
• Open Source Libraries: Transformers, PyTorch, and other dependencies
• Contributors: Everyone who has helped expand the cultural database

📞 Support

• Issues: GitHub Issues
• Discussions: GitHub Discussions
• Documentation: This README and inline code comments

---

Made with ❤️ for the ComfyUI community