Nano Banana - Image Processor

A flexible tool that uses Google's Gemini API to process images with customizable prompts. Perfect for batch processing slides, photos, and other images with AI-powered transformations.

Example: Slide Extraction

Transform photos of slides into clean, readable images automatically:

Before	After

The slide-extractor prompt automatically:

• Detects and extracts the slide from the photo
• Corrects perspective distortion
• Enhances contrast and readability
• Removes background clutter

Features

• Processes images through Google Gemini AI with customizable prompts
• Multiple prompt templates for different processing tasks
• Automatically renames files based on EXIF date metadata
• Batch processing of multiple images
• Easy prompt switching via command-line arguments

Prerequisites

• Node.js (v16 or higher recommended)
• pnpm package manager
• exiftool (for EXIF date extraction)
• Google Gemini API key

Getting Your Gemini API Key

1. Go to Google AI Studio
2. Sign in with your Google account
3. Click "Get API Key" or "Create API Key"
4. Copy your API key

The API is free for testing and light usage. Check Google's pricing page for current limits and rates.

Installation

1. Clone or download this repository

2. Install dependencies:

   pnpm install

3. Install exiftool (if not already installed):

   • macOS: brew install exiftool
   • Ubuntu/Debian: sudo apt-get install libimage-exiftool-perl
   • Windows: Download from exiftool.org

4. Create a .env file in the project root:

   echo "GEMINI_API_KEY=your_api_key_here" > .env

   Replace your_api_key_here with your actual Gemini API key.

Usage

Basic Workflow

1. Place your images in the input/ directory

2. Run the processing pipeline (uses default slide-extractor prompt):

   make all

This will:

• Rename files based on EXIF date (format: YYYY-MM-DD-HH-MM)
• Process each image with the selected prompt
• Save processed images to the output/ directory

Using Different Prompts

List available prompts:

make list-prompts

Process with a specific prompt:

make process PROMPT=slide-extractor

Or run the full pipeline with a custom prompt:

make all PROMPT=your-prompt-name

Individual Commands

Rename files only:

make rename

Process files only (without renaming):

make process

Install dependencies:

make install

Managing Prompts

Prompts are stored in the prompts/ directory as markdown files. Each prompt file name becomes its key.

Built-in Prompts:

• slide-extractor - Extracts slides from snapshots, corrects distortion, enhances readability

Creating Custom Prompts:

1. Create a new .md file in the prompts/ directory:

   nano prompts/my-custom-prompt.md

2. Write your prompt instructions in the file

3. Use it with:

   make process PROMPT=my-custom-prompt

Using nano-banana.js Directly

Process a single image with a prompt file:

node nano-banana.js --file path/to/image.jpg --prompt-file prompts/slide-extractor.md

Or with an inline prompt:

node nano-banana.js --file path/to/image.jpg --prompt "Your prompt here"

Specify custom output filename:

node nano-banana.js --file input.jpg --prompt-file prompts/slide-extractor.md --output custom-name.jpg

Project Structure

.
├── nano-banana.js      # Main processing script
├── Makefile           # Build automation
├── package.json       # Node.js dependencies
├── .env              # API key (not in git)
├── prompts/           # Prompt templates
│   └── slide-extractor.md  # Default slide extraction prompt
├── input/            # Place images here (contents ignored by git)
├── output/           # Processed images appear here (contents ignored by git)
└── docs/             # Documentation and example images
    ├── example-before.jpg
    ├── example-after.jpg
    ├── example-before-2.jpg
    ├── example-after-2.jpg
    ├── example-before-3.jpg
    └── example-after-3.jpg

Environment Variables

• GEMINI_API_KEY (required): Your Google Gemini API key

Troubleshooting

"GEMINI_API_KEY environment variable is not set"

• Ensure you've created a .env file with your API key
• The Makefile doesn't automatically load .env - you may need to source it: source .env && make process

"Could not find EXIF date for file"

• The image doesn't have EXIF metadata
• The file will be skipped during the rename process

"File is not an image"

• Ensure you're processing image files (JPEG, PNG, etc.)
• Check that the file isn't corrupted

License

This project uses the Google Gemini API, which has its own terms of service. Review Google's AI terms before use.