SORA2-Watermark-Remover

SORA2 Watermark Remover — AI-powered desktop application for video enhancement and visible watermark removal from Sora 2 generated content with TensorFlow/PyTorch neural inpainting, batch frame processing, CUDA GPU acceleration, and Tkinter GUI interface

╔══════════════════════════════════════════════════════════════════════════════╗
║   ███████╗ ██████╗ ██████╗  █████╗     ██╗   ██╗██████╗                    ║
║   ██╔════╝██╔═══██╗██╔══██╗██╔══██╗    ██║   ██║╚════██╗                   ║
║   ███████╗██║   ██║██████╔╝███████║    ██║   ██║ █████╔╝                   ║
║   ╚════██║██║   ██║██╔══██╗██╔══██║    ╚██╗ ██╔╝██╔═══╝                    ║
║   ███████║╚██████╔╝██║  ██║██║  ██║     ╚████╔╝ ███████╗                   ║
║   ╚══════╝ ╚═════╝ ╚═╝  ╚═╝╚═╝  ╚═╝      ╚═══╝  ╚══════╝                   ║
║   WATERMARK REMOVER · v1.1.2 · AI-Powered Video Processing                 ║
╚══════════════════════════════════════════════════════════════════════════════╝

AI-powered desktop application for video enhancement and visible watermark removal from SORA 2 generated content. Combines TensorFlow & PyTorch with a cyberpunk-themed GUI.

Features · Getting Started · Configuration · Usage · Project Structure · FAQ

---

Official Links

Resource	URL
OpenAI Sora	https://openai.com/sora
OpenAI Help Center (Sora)	https://help.openai.com/en/articles/12460853-creating-videos-with-sora
Sora System Card (PDF)	https://cdn.openai.com/pdf/50d5973c-c4ff-4d2d-986f-c72b5d0ff069/sora_2_system_card.pdf
C2PA Content Credentials	https://c2pa.org/
Repository	https://github.com/timanmoh/sora2-watermark-remover-gui

---

Features

Feature Status Modern Dark UI (Cyberpunk theme) ✓ Batch Processing (multi-file queue) ✓ Real-time Progress & ETA ✓ Dual AI Framework (TensorFlow + PyTorch) ✓ CNN-based Watermark Detection ✓ Temporal Frame Analysis ✓ SORA Signature Validation ✓

Feature Status ROI Region-of-Interest Optimization ✓ Neural Inpainting (OpenCV fallback) ✓ Configurable Quality Presets ✓ Output Formats (MP4, AVI, MOV, MKV) ✓ GPU Acceleration (CUDA) ✓ Drag & Drop Support ✓ Detailed Logging ✓

---

Getting Started

Prerequisites

Component	Requirement
OS	Windows 10/11, Linux (Ubuntu 20.04+), macOS
Python	3.10+
GPU	NVIDIA + CUDA (recommended, optional)
RAM	8 GB min, 16 GB recommended
Storage	2 GB free

Installation

# Clone repository
git clone https://github.com/timanmoh/sora2-watermark-remover-gui.git
cd sora2-watermark-remover-gui

# Create virtual environment
python -m venv .venv
.venv\Scripts\activate          # Windows
# source .venv/bin/activate     # Linux / macOS

# Install dependencies
pip install -r requirements.txt

# Run application
python main.py

Optional GPU Support

pip install cupy-cuda11x==12.3.0   # CUDA 11.x
# pip install cupy-cuda12x==12.3.0 # CUDA 12.x

Dependency Table

Package	Version	Purpose
numpy	≥1.24.0	Numerical operations
opencv-python	≥4.8.0	Video I/O, inpainting
Pillow	≥10.0.0	Image processing
moviepy	≥1.0.3	Video composition
tensorflow	≥2.13.0	Neural inpainting (TF backend)
torch	≥2.0.0	CNN detection, GPU
torchvision	≥0.15.0	Transforms, models
pyyaml	≥6.0	Configuration
rich	≥13.0.0	CLI formatting
tqdm	≥4.65.0	Progress bars

---

Configuration

Edit config.yaml in the project root or use the Settings menu ([3]) from the CLI.

# SORA2-WATERMARK-REMOVER Configuration

detection:
  confidence_threshold: 0.75    # 0.0 - 1.0, detection sensitivity
  detection_method: cnn         # cnn | hybrid
  temporal_analysis: true       # frame-to-frame consistency
  signature_check: true        # SORA video validation

output:
  quality: High                 # Draft | Standard | High | Lossless
  format: MP4                   # MP4 | AVI | MOV | MKV

quality_presets:
  Draft: 5                      # Mbps, fast preview
  Standard: 15
  High: 30
  Lossless: null                # max preservation

Tip: Lower confidence_threshold (e.g. 0.6) increases detection sensitivity but may produce false positives. Use High or Lossless for best output quality.

---

Usage

CLI Menu

  Enter your choice [1/2/3/4/0]: 

  [1] Start application
  [2] Install dependencies
  [3] Settings
  [4] About
  [0] Exit

GUI Workflow

1. Run python main.py and select [1] Start application
2. Add videos via [+] Add Videos or Ctrl+O
3. Adjust threshold and quality in Settings
4. Click START PROCESSING or press F5

Interface Layout

┌─────────────────────────────────────────────────────────────────┐
│  SORA 2 Video Editor v1.1.2                                      │
├─────────────────────────────────────────────────────────────────┤
│  Video Files                                                     │
│  [+] Add  [-] Remove  [x] Clear                                  │
│  ─────────────────────────────────────────────────────────────  │
│  • video1.mp4  • video2.mp4  • video3.mp4                       │
├─────────────────────────────────────────────────────────────────┤
│  Settings  [====75%====]  [HIGH]  [MP4]                          │
├─────────────────────────────────────────────────────────────────┤
│  Progress  ▓▓▓▓▓▓▓░░░  45%  ETA 2:35                            │
├─────────────────────────────────────────────────────────────────┤
│              [ ▶ START PROCESSING ]                              │
└─────────────────────────────────────────────────────────────────┘

Shortcuts: Ctrl+O Add · Delete Remove · Ctrl+A Select all · F5 Start · Esc Cancel

---

Project Structure

SORA2-Watermark-Remover/
│
├── main.py                 # Entry point (CLI menu + app launcher)
├── config.yaml            # User configuration
├── requirements.txt       # Python dependencies
│
├── gui/
│   └── main_window.py     # Tkinter window, cyberpunk theme
│
├── core/
│   ├── processor.py       # Video pipeline (load, detect, inpaint, write)
│   ├── inpainting.py      # Neural inpainting (TF/PyTorch + OpenCV fallback)
│   └── validator.py       # Input validation
│
├── detection/
│   ├── detector.py        # CNN-based watermark detection
│   ├── signature.py       # SORA signature validation
│   └── temporal.py        # Frame-to-frame consistency analysis
│
└── utils/
    ├── file_handler.py    # File I/O helpers
    ├── gpu_manager.py     # GPU detection and management
    └── logger.py          # Logging setup

---

FAQ

App won't start

• Ensure Python 3.10+ is installed: python --version
• Install dependencies: pip install -r requirements.txt
• Verify Tkinter: python -c "import tkinter"
• On Linux, install: sudo apt install python3-tk

GPU not detected

• Update NVIDIA drivers to the latest version
• Verify CUDA: nvidia-smi
• Install CUDA-compatible packages (cupy-cuda11x or cupy-cuda12x)
• Check that PyTorch sees CUDA: python -c "import torch; print(torch.cuda.is_available())"

Slow processing

• Enable GPU in settings and ensure CUDA packages are installed
• Close other GPU-intensive applications
• Lower output quality preset (e.g. Standard instead of Lossless)
• Reduce video resolution before processing

Watermark not fully removed

• Lower confidence_threshold in config (e.g. 0.65) for higher sensitivity
• Enable temporal_analysis and signature_check for better detection
• Use Lossless quality preset for maximum preservation
• Note: C2PA metadata (invisible provenance) is not removed by this tool

What watermark types does this handle?

This tool targets visible on-frame watermarks (e.g. "Made with Sora" overlays) using CNN detection and inpainting. OpenAI Sora also embeds C2PA Content Credentials (invisible cryptographic metadata). C2PA is designed to be tamper-resistant and is not removed by this application.

Supported input formats

MP4, AVI, MOV, MKV, and other formats supported by OpenCV's VideoCapture. Ensure ffmpeg is available on the system for best compatibility.

Output location

Processed videos are saved to a processed/ subdirectory next to each input file, with the suffix _processed (e.g. video.mp4 → processed/video_processed.mp4).

---

Disclaimer

This project is provided for educational and research purposes only. It demonstrates computer vision techniques for watermark detection and video inpainting. Users are responsible for ensuring their use complies with:

• OpenAI's Terms of Service and usage policies
• Applicable copyright and licensing laws
• Ethical guidelines for synthetic media

Removing watermarks from content may violate terms of service or licensing agreements. This tool is not affiliated with, endorsed by, or connected to OpenAI. Use at your own risk.

---

If this project is helpful, consider starring the repository.

ETH: 0x4B28F1a3C5D7E9b0A6f24c8E1B3a5F7d9C0e25ad