AI Metadata Viewer & Extractor

A high-performance JavaFX desktop application designed to inspect and extract generation metadata across the AI image generation ecosystem. It provides instant extraction, deep ComfyUI node-graph inspection, and raw metadata inspection for artists and developers.

[!NOTE] Favorites Library, Metadata Scrubber, and Speed Sorter have moved to Latent Library. Metadata Viewer is now focused purely as a lightweight, fast, single-screen extractor utility.

---

📸 Interface

Single-screen extractor: drop an image and every generation field populates instantly.

Raw Metadata: full unparsed output for non-standard or unrecognized sources, with one-click copy.

---

🔐 Stateless, Portable & Private

Built as a lightweight, zero-setup utility for quick, on-demand inspection.

• 100% Offline / No Telemetry: No network calls, no analytics, no cloud dependency. Your images and prompts never leave your machine.
• Zero Setup: No database, no config files, no background service. Drop an image and inspect it — nothing is written to disk.
• Portable Build: Ships as a single self-contained executable/app-image with a bundled Java 21 runtime. No system-wide Java installation required.
• Lightweight: Programmatic JavaFX (no FXML) keeps startup instant and the footprint small.

---

✨ Key Features

• Universal Compatibility: Intelligent parsing for ComfyUI (API & Workflow graphs), SwarmUI, Automatic1111 / Forge, InvokeAI, and NovelAI.
• Advanced ComfyUI Engine:
  • Graph Traversal: Resolves connected prompts, samplers, schedulers, models, and LoRAs across complex execution flows.
  • Custom Node Support: Recursively extracts parameters from advanced custom nodes (Power LoRA Loaders, KSampler variations, reroutes).
  • Extended Field Discovery: Surfaces Scheduler, Denoise, and Hires. fix as dedicated fields; Distilled CFG (flux-guidance) is folded into the CFG field. Model Hash and ControlNet are A1111/Forge-only today — still captured, but only visible via the Raw Metadata inspector for other sources.
• Physical Fallback: Reads physical image headers to guarantee valid image dimensions and file size even when metadata is missing or malformed.
• Latent Design System: Clean, modern dark UI styled against the Latent ecosystem's token specifications — shadows, gradients, and type scale included.
• Interactive UI:
  • Drag & Drop: Drop any PNG, JPG, JPEG, or WEBP directly onto the image preview for instant extraction — even to replace an image that's already loaded.
  • Fullscreen Preview: Click any image for a modal, high-res inspection view.
  • Raw Inspector: Debug non-standard outputs with a raw text/JSON inspection modal, with one-click copy.
  • Copy Feedback: Prompt and raw-metadata copy actions confirm with a toast instead of a silent clipboard write.
  • Responsive Layout: The metadata panel reflows underneath the image on narrower windows instead of clipping, and stat cards wrap to fit. Prompt boxes grow to fit their content, up to a cap, then scroll internally.
• Lightweight & Cross-Platform: Programmatic JavaFX (No FXML) with zero startup lag and cross-platform build support.

---

💻 System Requirements

• OS: Windows 10/11 (64-bit), Linux, or macOS (11+).
• Memory: Minimal. A few hundred MB of RAM is sufficient — no database or background indexing.
• Storage: ~100MB for the application (bundled Java 21 runtime included).
• Network: Not required. Metadata Viewer performs no network calls of any kind.

---

🛠️ Technical Architecture

• Strategy Pattern: Tool-specific parsing strategies (ComfyUIStrategy, CommonStrategy, SwarmUIStrategy, InvokeAIStrategy, NovelAIStrategy) with heuristic chunk scoring.
• Stateless & Portable: Zero background database or config locking — drop and inspect with zero setup overhead.
• Technology Stack: Java 21, JavaFX 21, Jackson, metadata-extractor (Drew Noakes), Ikonli (FontAwesome), SLF4J, JUnit 5.

---

🚀 Getting Started

1. Download the appropriate file for your OS:
   • Windows: MetadataViewer-windows-X.X.X.exe — single self-contained file, no installation required.
   • Linux: MetadataViewer-linux-X.X.X.zip
   • macOS: MetadataViewer-macos-X.X.X.zip
2. Extract the .zip (macOS/Linux only — a real "Extract All", not just browsing into the archive) and run the app from inside the extracted folder. No installation is required.
3. Drop an image onto the preview to extract its generation metadata instantly.

🍎 macOS Users: Because this app is not yet signed with an Apple Developer Certificate, you may see an error saying the app is "damaged and can't be opened." This is a standard macOS security message for unsigned apps.

To fix this:

1. Move the app to your Applications folder.
2. Open Terminal.
3. Run the following command to clear the quarantine attribute:

   sudo xattr -cr "/Applications/MetadataViewer.app"
   

4. You can now open the app normally.

Building from source instead? See CONTRIBUTING.md.

---

📜 License

Distributed under the MIT License with the Commons Clause. Free to use, modify, and share. The Commons Clause restricts selling the Software or offering it as a paid hosted/consulting service. See LICENSE.md for the full text.

---

💖 Support the Project

If AI Metadata Viewer has streamlined your workflow, consider supporting its ongoing development.

---

Developed by

Copyright (c) 2025-2026 Alexander Nilsson