English | Русский

ComfyUI Nano Banana Node

This repository contains a custom node for ComfyUI, "Nano banana", designed for advanced image editing using the Google Gemini API.

This project is based on the original Visionatrix/ComfyUI-Gemini repository.

Features

The "Nano banana" node allows you to perform context-aware image editing by providing:

• A base image to be modified.
• One or more reference images (for style, patterns, etc.).
• A descriptive text prompt detailing the desired changes.

How to Use

1. image_to_edit: Connect the primary image you want to modify to this input. The output image will maintain the aspect ratio of this input.
2. reference_image_1 / reference_image_2 / reference_image_3 / reference_image_4 / reference_image_5: Use these inputs for any reference images, such as textures, patterns, or style examples that you want to apply to the main image.
3. prompt: Write a descriptive, narrative prompt explaining the edit. For example: "On the main image, change the color of the walls to a rich red, using the texture from the reference image."

Outputs

• image: The resulting edited image. In case of an error, a black image will be outputted to prevent workflow crashes.
• status: A string indicating the status of the generation:
  • "Complete": The image generation was successful.
  • "Error: Prompt or image was blocked by safety filters.": The Gemini API refused to generate content due to safety settings.
  • "Error: Could not decode image from API response: [details]": The API returned data that could not be interpreted as an image.
  • "Error: [other details]": A general error occurred during processing or API communication.