Image Description Toolkit
AI-powered batch image description tool supporting multiple vision models (Ollama, OpenAI GPT-4o, Claude — through the API or through your Claude subscription).
Quick Start
Installation
Download from the releases page:
| File | Platform |
|---|
ImageDescriptionToolkitSetup-4.8.0-windows.exe | Windows 10/11, 64-bit |
IDT-4.8.0-macos-arm64.dmg | macOS, Apple Silicon |
Either one gives you all the applications:
- idt - Command-line interface for batch processing and automation
- ImageDescriber - GUI with integrated viewer, prompt editor, and configuration manager
- IDT Chat - An accessible chat client for Ollama, Claude, OpenAI, Claude Code and Apple Intelligence. Not an image tool: it is a general-purpose chat application built for keyboard and screen reader use.
One install covers them all; you never update them separately.
Standalone builds are also published if you want a single tool without an installer:
idt-4.8.0-windows-x64.exe, ImageDescriber-4.8.0-windows-x64.exe,
IDTChat-4.8.0-windows-x64.exe, and idt-4.8.0-macos-arm64.tar.gz.
SHA256SUMS.txt lets you verify any download.
No Python required. The Windows installer is signed; the macOS build is signed and
notarized.
Latest Release (v4.8.0): Windows AI, a new provider that describes images on a Copilot+ PC with Windows' own model: on the PC itself, with no API key, account or cost.
Quick Start
GUI (Easiest for Most Users)
- Install Image Description Toolkit from the installer
- Launch
imagedescriber.exe from the install folder (default: C:\IDT\)
- Choose a directory of images (File → Load Directory)
- Select your AI provider, model, and prompt style in the interface
- Process all images (Processing → Process All Undescribed)
Your images will be described automatically! Use Tools → Edit Prompts to customize description styles, or File → Switch to Viewer to browse results.
Command Line (For Batch Processing)
- Open a command prompt
- Change to the install directory:
cd C:\IDT
- Run the interactive guide:
idt guideme
- Answer the prompts to configure and run your workflow
Results are saved in the Descriptions/ folder with an HTML viewer.
Advanced Usage
# Describe a folder of images
idt describe path/to/images
# Check whether a newer version is available
idt update
Features
- Two Powerful Applications: GUI for visual workflow, CLI for automation
- Multiple AI Providers: Ollama (local), OpenAI, Claude (API key), Claude Code (your Claude Pro/Max subscription, no API key), Apple Intelligence (on-device on macOS 27, no API key), Windows AI (on-device on Copilot+ PCs, no API key)
- Batch Processing: Process directories of images automatically
- Video Frame Extraction: Extract and describe frames from videos
- Integrated Viewer: Browse and monitor workflows in real-time (built into GUI)
- Integrated Tools: Prompt editor and configuration manager built into GUI
- Workflow Management: Organized results with metadata tracking
- Re-describe Feature: Test different models/prompts on same images
- Analysis Tools: Compare models, review content, export to CSV/Excel
Documentation
User Guides
Configuration & Setup
Developer Documentation
Requirements
- Pre-built Executable: No Python required - download installer from releases
- AI Provider (choose one or more):
macOS
- Apple Silicon — the published build is arm64 only. Intel Macs need to build from source.
- No Python required for the pre-built app
- Installation: Download
IDT-{version}-macos-arm64.dmg from releases, open it, and drag ImageDescriber to Applications
- Build from Source: See macOS Build Guide
- AI Providers: Same as Windows (Ollama, OpenAI, Claude, Claude Code), plus Apple Intelligence on macOS 27 and MLX on Apple Silicon
Development (All Platforms)
- Python 3.10+ (for development only, not required for built executables)
- AI Provider (choose one or more):
- Ollama (local, free)
- OpenAI API key
- Anthropic Claude API key
Support
License
MIT License - see LICENSE file for details.
