CleanShot Pro v2.3
Author: imsecure4@gmail.com
Status: Research / Privacy Tool
License: Free for personal use – attribution required
CleanShot Pro is an advanced image preprocessing tool designed to
disrupt AI-based fingerprinting while keeping the image visually identical
to the human eye.
It adds minimal adversarial perturbations, removes all metadata,
and applies subtle transformations that degrade AI detectors without
destroying image quality.
Weak but strategically placed noise to confuse ML models.
AI detectors rely heavily on mid-band patterns.
Maintains natural colors — no ugly distortions.
Output PNG contains zero EXIF or camera fingerprints.
light→ subtle anti-AImedium→ balancedstrong→ maximum AI fingerprint disruption
Process entire folders using all CPU cores.
No pywt or deep-learning dependencies.
Install dependencies with:
pip install pillow numpy
Python version:
Python 3.7+ (Linux, macOS, Windows)
---
🚀 Usage
Process a single image:
python3 cleanshot_pro.py input.jpg
Choose strength:
python3 cleanshot_pro.py input.jpg --preset medium
python3 cleanshot_pro.py input.jpg --preset strong
Process an entire folder:
python3 cleanshot_pro.py images/ --preset strong --out cleaned/
Use all CPU cores:
python3 cleanshot_pro.py images/ -t 0
Custom output directory:
python3 cleanshot_pro.py input.jpg --out protected/
---
🗂️ Output
All processed images are saved as:
cleanshot_<preset>_<original_name>.png
Examples:
cleanshot_light_photo1.png
cleanshot_strong_profile.png
These PNGs contain:
No EXIF
No device metadata
No camera information
No GPS
Modified luminance fingerprint
Distorted mid-frequency signature
---
⚠️ Disclaimer
This project is created strictly for:
privacy research
anti-fingerprinting experiments
understanding AI detection models
academic adversarial ML research
You are fully responsible for any misuse.
By using this tool, you agree:
I 'imsecure4'am not responsible for any illegal or harmful usage
You must include attribution (“Created by : imsecure4@gmail.com")
if you redistribute, modify, or extend the tool
---
🤝 Contribution
You may modify or improve the tool only if you credit the original author:
Original Author: imsecure4@gmail.com
Pull requests for improvements are welcome.
---
⭐ Future Planned Features
Adaptive per-image adversarial search
Face-recognition targeted evasion
Deep gradient-free perturbation engine
Multi-model resistance mode
---
📧 Contact
For collaboration, research use, or improvements:
imsecure4@gmail.com