CalvinBackup/reverse-SynthID
1
0
Fork 0
mirror of https://github.com/aloshdenny/reverse-SynthID.git synced 2026-05-08 06:06:41 +02:00
Code Issues Packages Projects Releases Wiki Activity

watermark investigation

Browse Source
This commit is contained in:
Alosh Denny
2025-12-16 16:34:03 +05:30
parent 21b094474e
commit e02b4a11da
43 changed files with 7322 additions and 21 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+177 -21
View File
@@ -2,24 +2,119 @@
<img src="assets/synthid-watermark.jpeg" alt="SynthID Watermark Analysis" width="100%">
</p>
<h1 align="center">🔍 SynthID Watermark Reverse Engineering</h1>
<h1 align="center">🔍 AI Watermark Reverse Engineering</h1>
<p align="center">
<b>Discovering Google's hidden AI watermark patterns through signal analysis</b>
<b>Discovering hidden AI watermark patterns through signal analysis</b>
</p>
<p align="center">
<img src="https://img.shields.io/badge/Python-3.10+-blue?style=flat-square&logo=python" alt="Python">
<img src="https://img.shields.io/badge/License-Research-green?style=flat-square" alt="License">
<img src="https://img.shields.io/badge/Status-Complete-success?style=flat-square" alt="Status">
<img src="https://img.shields.io/badge/Accuracy-84%25-brightgreen?style=flat-square" alt="Accuracy">
<img src="https://img.shields.io/badge/Images_Analyzed-123,268-brightgreen?style=flat-square" alt="Images">
<img src="https://img.shields.io/badge/Detection_Rate-99.9%25-success?style=flat-square" alt="Detection">
</p>
---
## 🎯 Overview
This project reverse-engineers **Google's SynthID watermarking technology** by analyzing 250 AI-generated images from Gemini. Since the neural network encoder/decoder is proprietary, we use signal processing techniques to discover the watermark's structure.
This project reverse-engineers **AI watermarking technologies** by analyzing AI-generated and AI-edited images. We use signal processing techniques to discover watermark structures without access to proprietary neural network encoders/decoders.
### Projects
| Analysis | Images | Detection Rate | Key Finding |
|:---------|:------:|:--------------:|:------------|
| **[Nano-150k Investigation](#-nano-150k-watermark-investigation)** | 123,268 | 99.9% | Multi-layer frequency + spatial watermarking |
| **[SynthID Analysis](#-synthid-google-gemini-analysis)** | 250 | 84% | Spread-spectrum phase encoding |
---
## 🔬 Nano-150k Watermark Investigation
Analysis of **123,268 AI-edited image pairs** from the Nano-150k dataset to detect and characterize embedded watermarks.
### Key Discovery
AI-edited images contain **multi-layer watermarks** using both frequency domain (DCT/DFT) and spatial domain (color shifts) embedding techniques. The watermarks are invisible to humans but detectable via statistical analysis.
### Detection Results
| Metric | Rate | Description |
|:-------|:----:|:------------|
| **Frequency Domain Modifications** | 100.0% | All images show spectral changes |
| **Significant Color Shifts** | 95.3% | Mean shift > 1.0 in RGB channels |
| **Perceptual Hash Changes** | 66.0% | Invisible modifications detected |
| **LSB Anomalies** | 10.2% | Least significant bit patterns |
| **2+ Watermark Indicators** | 99.9% | Multi-layer evidence |
| **3+ Watermark Indicators** | 69.2% | Strong multi-layer evidence |
### Watermark Confidence Distribution
```
0 indicators: 0 ( 0.0%)
1 indicator: 122 ( 0.1%)
2 indicators: 37,832 (30.7%) ███████████████
3 indicators: 74,525 (60.5%) ██████████████████████████████
4 indicators: 10,789 ( 8.8%) ████
```
### Extracted Watermark Visualizations
<table>
<tr>
<td width="50%">
**Extracted Watermark Pattern**
<img src="watermark_investigation/WATERMARK_EXTRACTED.png" width="100%">
</td>
<td width="50%">
**Comprehensive Analysis**
<img src="watermark_investigation/WATERMARK_FINAL_ANALYSIS.png" width="100%">
</td>
</tr>
<tr>
<td width="50%">
**Frequency Spectrum**
<img src="watermark_investigation/WATERMARK_frequency_spectrum.png" width="100%">
</td>
<td width="50%">
**Enhanced Difference Pattern**
<img src="watermark_investigation/WATERMARK_enhanced_difference.png" width="100%">
</td>
</tr>
</table>
### Analysis by Edit Category
| Category | Image Pairs | Avg Freq Diff | Watermark Strength |
|:---------|:-----------:|:-------------:|:------------------:|
| hairstyle | 16,012 | 1.786 | High |
| sweet_headshot | 16,008 | 1.759 | High |
| black_headshot | 17,700 | 1.735 | High |
| background | 32,765 | 1.037 | Medium |
| time-change | 18,178 | 1.028 | Medium |
| action | 22,605 | 1.013 | Medium |
### Processing Statistics
- **Total Processing Time**: 170.2 minutes
- **Processing Rate**: 12.1 pairs/second
- **Success Rate**: 100% (0 failed loads)
---
## 🔬 SynthID (Google Gemini) Analysis
Analysis of **250 AI-generated images** from Google Gemini to reverse-engineer SynthID watermarking.
### Key Discovery
@@ -77,10 +172,20 @@ SynthID uses **spread-spectrum phase encoding** in the frequency domain—not LS
## 📁 Project Structure
```
synthid-demarker/
reverse-SynthID/
├── 📄 README.md # This file
├── 📋 requirements.txt # Python dependencies
│
├── 🔍 watermark_investigation/ # Nano-150k Analysis (NEW)
│ ├── WATERMARK_EXTRACTED.png # Final extracted watermark
│ ├── WATERMARK_FINAL_ANALYSIS.png # Comprehensive visualization
│ ├── WATERMARK_enhanced_difference.png # Enhanced pattern
│ ├── WATERMARK_frequency_spectrum.png # Frequency domain
│ ├── WATERMARK_signed_pattern.png # Signed watermark
│ ├── watermark_FULL_123k_results.json # Complete results
│ ├── watermark_evidence/ # Visual evidence
│ └── *.py # Analysis scripts
│
├── 💻 src/
│ ├── analysis/
│ │ ├── synthid_codebook_finder.py # Pattern discovery
@@ -109,8 +214,8 @@ synthid-demarker/
### Installation
```bash
git clone https://github.com/yourusername/synthid-demarker.git
cd synthid-demarker
git clone https://github.com/yourusername/reverse-SynthID.git
cd reverse-SynthID
# Create virtual environment
python -m venv venv
@@ -120,7 +225,20 @@ source venv/bin/activate # Windows: venv\Scripts\activate
pip install -r requirements.txt
```
### Detect Watermark
### Run Nano-150k Watermark Analysis
```bash
# Full analysis on all 123k pairs (takes ~3 hours)
python watermark_investigation/watermark_full_123k_analysis.py
# Extract final watermark visualization
python watermark_investigation/extract_final_watermark.py
# Quick sample analysis (1000 pairs)
python watermark_investigation/watermark_full_analysis.py
```
### Detect SynthID Watermark
```bash
python src/extraction/synthid_codebook_extractor.py detect "path/to/image.png" \
@@ -156,30 +274,68 @@ python src/analysis/deep_synthid_analysis.py
## 🧠 How It Works
### 1. Pattern Discovery
Analyze noise patterns across multiple images to find consistent structures that persist despite varying image content.
### Nano-150k Watermark Detection
### 2. Frequency Analysis
Use FFT to identify carrier frequencies where the watermark is embedded through phase modulation.
1. **Frequency Domain Analysis**: Compute FFT differences between original and edited images
2. **LSB Pattern Detection**: Analyze least significant bit distributions for anomalies
3. **Color Shift Measurement**: Detect systematic RGB channel modifications
4. **Perceptual Hashing**: Compare perceptual hashes to find invisible changes
5. **Multi-Indicator Scoring**: Combine multiple detection methods for confidence
### 3. Phase Coherence
Measure phase consistency at carrier frequencies—high coherence indicates watermark presence.
### SynthID Detection
### 4. Codebook Extraction
Build reference patterns from averaged signals across many watermarked images.
### 5. Detection
Compare test image against codebook using correlation, phase matching, and structure ratio metrics.
1. **Pattern Discovery**: Analyze noise patterns across multiple images to find consistent structures
2. **Frequency Analysis**: Use FFT to identify carrier frequencies with phase modulation
3. **Phase Coherence**: Measure phase consistency at carrier frequencies
4. **Codebook Extraction**: Build reference patterns from averaged signals
5. **Detection**: Compare test image against codebook using correlation metrics
## 📊 Technical Details
### Watermark Characteristics
### Nano-150k Watermark Characteristics
- **Embedding Domains**: Frequency (DCT/DFT) + Spatial (color shifts)
- **Detection Methods**: FFT analysis, LSB statistics, perceptual hashing
- **Signal Strength**: Mean freq diff ~1.32, color shifts 32-35 pixel values
- **Robustness**: Survives JPEG compression, consistent across edit types
- **Categories Analyzed**: background, action, time-change, headshot, hairstyle
### SynthID Watermark Characteristics
- **Embedding Domain**: Frequency (FFT phase)
- **Signal Strength**: ~0.1-0.15 pixel values
- **Carrier Count**: 100+ frequency locations
- **Robustness**: Survives moderate compression
### Detection Algorithm
### Detection Algorithms
**Nano-150k Multi-Indicator Detection:**
```python
def detect_watermark(original, edited):
indicators = 0
# 1. Frequency domain analysis
freq_diff = compute_fft_difference(original, edited)
if freq_diff > 0.5:
indicators += 1
# 2. Color shift detection
color_shift = compute_color_shift(original, edited)
if any(abs(shift) > 1.0 for shift in color_shift):
indicators += 1
# 3. LSB anomaly detection
lsb_deviation = compute_lsb_deviation(edited)
if any(dev > 0.02 for dev in lsb_deviation):
indicators += 1
# 4. Perceptual hash comparison
phash_dist = compute_phash_distance(original, edited)
if 5 < phash_dist <= 30:
indicators += 1
return indicators >= 2, indicators
```
**SynthID Detection:**
```python
def detect_synthid(image, codebook):
# 1. Extract noise pattern