Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings
Open more actions menu

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
6 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Python OpenCV MIT Stars

👻 Ghost Font Decoder

Reveal hidden messages from anti-AI motion fonts — the tool that reads what AI can't.


Ghost Font (mixfont.com/ghost-font) is a new typeface designed to be unreadable by AI. It hides text as moving dots — invisible in any single frame, invisible to OCR, invisible to multimodal models like GPT-5 and Claude 4.

This decoder tracks those dots across all frames, rebuilds the motion trails, and reveals the hidden message. It even attempted OCR to read the result automatically.

Human-readable → ✅
AI-readable → only with this tool


✨ Features

🎥 Drop any Ghost Font video MP4, AVI, MOV, MKV
🔍 3 detection methods Transition-tracking, dark-count, frame-diff
🤖 Auto OCR EasyOCR reads the decoded image (best-effort)
📄 Export PDF (image + text), TXT, PNG
🎛️ Tunable Sensitivity (percentile), dilation, method
🎨 Dark UI Modern Tkinter GUI with drag-drop

🚀 Quick Start

git clone https://github.com/roohan-514/ghost-decoder.git
cd ghost-decoder
pip install opencv-python numpy easyocr fpdf2 pillow
python main.py

🖥️ CLI

python cli.py video.mp4 -o result.png
python cli.py video.mp4 --pdf result.pdf
python cli.py video.mp4 -m transition -t 99 -b 8

🧠 How It Works

Video frames  →  detect dark moving dots  →  count transitions per pixel
  →  keep top 1% most-active pixels  →  dilate into letter shapes  →  OCR
  1. Dot detection — finds dark moving pixels (value < 50) in each frame
  2. Transition counting — tracks light→dark transitions per pixel; moving dots cause frequent transitions while static decoys don't
  3. Percentile filtering — keeps only the most-active pixels where letter paths are traced
  4. Dilation — connects nearby dots into readable strokes
  5. OCR (optional) — EasyOCR attempts to read the final image

⚠️ Why This Matters

Ghost Font is part of a growing trend: anti-AI content designed to be readable only by humans. As AI gets better at OCR, vision, and understanding images, projects like Ghost Font fight back — and tools like this decoder exist at the boundary.

This is a cat-and-mouse game:

  • AI generates undreadable fonts → humans build decoders
  • AI improves → new anti-AI techniques emerge

We're building the tools that keep human communication human.

📸 Try It Yourself

  1. Go to mixfont.com/ghost-font
  2. Type a secret message
  3. Download the video
  4. Drop it into the decoder

🤝 Contributing

PRs welcome! Ideas:

  • Better OCR preprocessing for ghost dots
  • Video-native model integration
  • A web version (WASM + OpenCV.js)
  • Support for other anti-AI fonts

📜 License

MIT


If AI won't read it — we will.
Star on GitHub · Try Ghost Font

About

Signal analysis toolkit for detecting and decoding hidden patterns and encrypted data in various signal formats

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

Morty Proxy This is a proxified and sanitized view of the page, visit original site.