ComfyUI_Nano_Banana

A set of custom nodes for ComfyUI that leverage the Gemini 2.5 Flash Image Preview API to generate images from text prompts, single images, and multiple images.

Installation

1. Clone this repository into your custom_nodes folder.

   cd ComfyUI/custom_nodes
   git clone https://github.com/ru4ls/ComfyUI_Nano_Banana.git
   

2. Install the required dependencies:

   pip install -r ComfyUI_Nano_Banana/requirements.txt
   

Nodes

Nano Banana

Model Limitations

Please be aware that the gemini-2.5-flash-image-preview model used by these nodes has a fixed output resolution of 1024x1024 pixels. While the node provides width and height inputs, these are primarily for compatibility with ComfyUI's interface and do not directly control the output resolution of the generated image. The model will always produce a 1024x1024 image.

Attempting to influence the aspect ratio by including it in the prompt (e.g., "a dog, aspect ratio 16:9") acts as a compositional hint to the model, guiding how it frames the image within the 1024x1024 canvas. It does not change the actual pixel dimensions of the output.

This node provides a flexible interface for image generation, supporting text-to-image and image-to-image workflows with up to five reference images.

Inputs:

• prompt (STRING): The text prompt for image generation or manipulation.
• seed (INT, optional): Seed for reproducibility (default: 0).
• width (INT, optional): Width of the generated image (default: 1024). Used primarily for text-to-image generation.
• height (INT, optional): Height of the generated image (default: 1024). Used primarily for text-to-image generation.
• image_1 to image_5 (IMAGE, optional): Up to five reference images. Provide at least one image for image-to-image generation.

Outputs:

• image (IMAGE): The generated image.

API Key Setup

To use these nodes, you need a Google AI API key. You can obtain one from the Google AI Studio. Please note that the Gemini API is a paid service and may incur costs.

Create a .env file in the ComfyUI_Nano_Banana directory with the following content:

REPLICATE_API_KEY="YOUR_API_KEY"

Example Usage

!Important Make sure your API key is set up in the .env file.

Text to Image Generation

1. Add the NanoBanana node to your workflow.
2. Enter a prompt.
3. Ensure no image_ inputs are connected.
4. Connect the output image to a PreviewImage node to see the result.

Sample Prompt: "A delicious-looking cheeseburger with fresh lettuce, tomatoes, and melted cheese on a sesame seed bun, professional food photography"

Image to Image Generation (with 1 to 5 reference images)

1. Add the NanoBanana node to your workflow.
2. Connect one or more LoadImage nodes (up to 5) to the image_1 to image_5 inputs.
3. Enter a prompt describing the desired changes or outcome.
4. Connect the output image to a PreviewImage node.

Sample Prompt: "Image_2 serves Image_1 to Image_3 in a small, classic fisherman's bar."

License

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