MCP Image Zoom-In Tool

A Model Context Protocol (MCP) server that provides image zoom-in capabilities for AI models.

Features

• Image Cropping: Crop specific regions from images using bounding box coordinates
• Smart Resizing: Automatically resize cropped images for Vision Transformer compatibility
• Multiple Input Sources: Support for URLs, local files, and base64 encoded images
• Relative Coordinates: Use normalized [0, 1000] coordinate system

Installation

# Using uvx (recommended for MCP)
uvx mcp-image-zoom

# Using pip
pip install mcp-image-zoom

Usage

As MCP Server

Run the server directly:

mcp-image-zoom

MCP Configuration

Add to your MCP client configuration:

{
  "mcpServers": {
    "image-zoom": {
      "command": "uvx",
      "args": ["mcp-image-zoom"]
    }
  }
}

Available Tools

image_zoom_in

Zoom in on a specific region of an image.

Parameters:

• image_source (string): Image URL, file path, or base64 string
• bbox_x1 (number): Left edge (0-1000)
• bbox_y1 (number): Top edge (0-1000)
• bbox_x2 (number): Right edge (0-1000)
• bbox_y2 (number): Bottom edge (0-1000)
• label (string, optional): Region label

Returns:

• cropped_image: Base64 encoded cropped image
• original_size: [width, height]
• crop_region: [x1, y1, x2, y2] in pixels
• output_size: [width, height] of output

get_image_info

Get basic information about an image.

Parameters:

• image_source (string): Image URL, file path, or base64 string

Returns:

• width, height: Image dimensions
• format: Image format
• mode: Color mode

Coordinate System

The tool uses relative coordinates [0, 1000]:

• (0, 0) = top-left corner
• (1000, 1000) = bottom-right corner

Example: To crop the center-right region:

{
  "bbox_x1": 500,
  "bbox_y1": 250,
  "bbox_x2": 1000,
  "bbox_y2": 750
}

License

Apache-2.0