ComfyUI-Imagen-Gemini

A custom node for ComfyUI that integrates Google's latest Imagen 4 and Imagen 3 generative models via the Gemini API.

Generate high-quality images directly within your ComfyUI workflows using Google's state-of-the-art diffusion models. Note that: The "allow_all" parameter value is not allowed in EU, UK, CH, MENA locations.

✨ Features

• Model Support: Access to the latest models including:
  • Imagen 4: Ultra, Standard, and Fast variants.
  • Imagen 3: Standard and Fast variants.
• Resolution Control: Select between 1K and 2K native generation.
• Aspect Ratios: Native support for 1:1, 3:4, 4:3, 9:16, and 16:9.
• Safety Settings: Configurable person generation filters (allow_adult, allow_all, dont_allow).
• Guidance Output: Includes a secondary text output containing official prompting guidance for Imagen 4.
• Secure Auth: Supports API keys via node widget or environment variables (GEMINI_API_KEY).

🛠️ Installation

Option 1: Git Clone (Manual)

1. Navigate to your ComfyUI custom_nodes directory.

2. Clone this repository:

   git clone [https://github.com/yourusername/ComfyUI-Imagen-Gemini.git](https://github.com/yourusername/ComfyUI-Imagen-Gemini.git)
   

3. Install dependencies (if necessary):

   pip install -r requirements.txt
   

   (Note: This node primarily uses standard libraries like requests, torch, and Pillow which are likely already installed in your ComfyUI environment.)

4. Restart ComfyUI.

🔑 API Key Configuration

You need a Google Gemini API key to use this node. You can get one here: Google AI Studio.

You can provide the key in two ways:

1. Environment Variable (Recommended): Set a GEMINI_API_KEY environment variable on your system. The node will automatically detect this, allowing you to leave the API key field in the UI blank.

2. Node Widget: Paste your API key directly into the gemini_api_key field on the node.

🎛️ Usage

1. Add Node: Right-click > ImagenGemini > Imagen Gemini.
2. Connect Prompt: Connect a string primitive or write your prompt in the widget.
3. Outputs:
   • IMAGE: The generated image tensor (connect to PreviewImage or SaveImage).
   • GUIDANCE: A string output containing prompting tips specific to Imagen models (connect to ShowText or similar).

Parameters

Parameter	Description
prompt	The positive text description for image generation.
model	Select the model version (e.g., imagen-4.0-generate-001, imagen-3.0-fast-generate-001).
gemini_api_key	Your Google API Key. Leave empty if using env var.
aspect_ratio	The shape of the image (1:1, 16:9, etc.).
resolution	1K or 2K generation.
num_images	Batch size (1 to 4 images per request).
person_generation	Controls generation of people: allow_adult (default), allow_all, or dont_allow.

📋 Requirements

• ComfyUI
• Python 3.x
• requests
• torch
• Pillow
• numpy

⚠️ Notes

• Negative Prompts: Google's Imagen API implementation for these models does not currently support negative prompts or system instructions; therefore, the node only accepts a positive prompt.
• API Limits: Usage is subject to Google Gemini API quotas and pricing. Please check your Google AI Studio dashboard for details.

License

MIT