ComfyUI-NanoBanano

ComfyUI-NanoBanano
★ 75

多模态图像生成图像编辑角色一致性
基于Google Gemini 2.5（Nano Banana）的ComfyUI自定义节点，支持多模态图像生成、编辑与风格迁移，强调角色一致性与批量处理。
💡 用文本与最多5张参考图批量生成或编辑并保持角色一致性。
🍴 18 Forks💻 Python🔄 2025-08-27
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/8f9eee5e2cdb
📦 requirements.txt
google-generativeai>=0.8.0
torch>=1.11.0
pillow>=8.3.0
numpy>=1.21.0
requests>=2.25.0
📄 README

ComfyUI-NanoBanano

A ComfyUI custom node for Google’s Gemini 2.5 Flash Image (aka “Nano Banana”) model – the state-of-the-art image generation and editing AI.

Features

  • Multi-Modal Operations: Generate, edit, style transfer, and object insertion
  • Up to 5 Reference Images: Support for complex multi-image operations
  • Character Consistency: Maintain identity across edits and generations
  • Batch Processing: Generate up to 4 images per request
  • Quality Control: Temperature and quality settings
  • Aspect Ratio Support: Multiple format options (1:1, 16:9, 9:16, 4:3, 3:4)
  • Cost Tracking: Built-in cost estimation (~$0.039 per image)

    • Requirements

  • ComfyUI
  • Paid Google Gemini API Key (Free tier does not support image generation)
  • Python packages (installed automatically):
  • google-generativeai
  • torch
  • pillow
  • numpy
  • requests

    • Installation

    Method 1: Git Clone (Recommended)

    cd ComfyUI/custom_nodes/
git clone https://github.com/ShmuelRonen/ComfyUI-NanoBanano.git
cd ComfyUI-NanoBanano
pip install -r requirements.txt

    API Key Setup

    1. Get Your API Key

  • Visit Google AI Studio
  • Sign in and enable billing (paid tier required)
  • Generate API key (starts with AIza...)

    • 2. Configure the Key

    Environment Variable (Recommended):

    export GEMINI_API_KEY="your_api_key_here"

    Or enter directly in the node’s api_key field

    Usage

  • Find the Node: Search “Nano Banana” in ComfyUI
  • Select Operation:
  • Generate: Create new images from text
  • Edit: Modify existing images
  • Style Transfer: Apply styles from references
  • Object Insertion: Add elements to scenes

  • Key Parameters:
  • prompt: Describe what you want
  • reference_image_1-5: Upload reference images
  • temperature: Creativity (0.0-1.0)
  • batch_count: Images per run (1-4)
  • aspect_ratio: Only affects generation, not editing

    • Examples

    Basic Generation

    Operation: generate
Prompt: "A dragon flying over a cyberpunk city at sunset"
Aspect Ratio: 16:9

    Image Editing

    Operation: edit  
Reference Image: [Your photo]
Prompt: "Add falling snow and winter atmosphere"

    Style Transfer

    Operation: style_transfer
Reference Image 1: [Content]
Reference Image 2: [Style reference]
Prompt: "Apply watercolor painting style"

    Important Limitations

  • Output Resolution: API limits to ~1024px max dimension
  • Cost: ~$0.039 per image generated
  • API Access: Requires paid Gemini subscription
  • Rate Limits: Vary by subscription tier

    • Troubleshooting

    “API key not valid”

  • Ensure billing is enabled in Google Cloud Console
  • Free tier cannot access image generation models

    • “No images found in response”

  • Try more explicit prompts: “Generate an image of…”
  • Check API rate limits and billing status

    • Module errors

    pip install google-generativeai pillow torch numpy requests

    Cost Information

  • Per Image: ~$0.039 USD
  • Batch of 4: ~$0.156 USD
  • Node displays cost estimates automatically

    • Contributing

  • Fork this repository
  • Create feature branch (git checkout -b feature/name)
  • Commit changes (git commit -m 'Add feature')
  • Push branch (git push origin feature/name)
  • Open Pull Request

    • License

    MIT License – see LICENSE file for details.

    Support

  • Issues: GitHub Issues
  • ComfyUI Community: Discord #custom-nodes channel

    • Note: Unofficial implementation. Google and Gemini are trademarks of Google LLC. Repository Structure