AI Video Watermark Removal SaaS

A full-stack SaaS platform for removing watermarks from AI-generated videos using DeMark-World.

🌟 Features

• Universal Watermark Removal: Supports Sora 2, Veo 3, Runway Gen-4, Kling 1.5/1.6, Luma Dream Machine, Pika 2, and more
• Multiple Quality Options: Fast (LaMa) and High Quality (E2FGVI_HQ) processing modes
• Modern UI: Beautiful drag-and-drop interface built with Next.js 15 and Tailwind CSS
• Scalable Architecture: FastAPI backend with Celery workers on GPU instances
• Cloud Storage: Direct uploads to Cloudflare R2 (S3-compatible)

📚 Documentation

• Deployment Guide - Step-by-step production deployment
• API Documentation - Complete API reference
• Admin Setup - How to grant admin privileges
• Contributing Guide - How to contribute to the project

🏗️ Architecture

Frontend (Next.js 15)
    ↓ Upload Request
Backend (FastAPI)
    ↓ Job Queue (Redis)
Worker (Celery on GPU)
    ↓ DeMark-World Processing
Storage (Cloudflare R2)

📁 Project Structure

/
├── frontend/         # Next.js application
├── backend/          # FastAPI server
│   ├── main.py       # API endpoints
│   ├── database.py   # SQLAlchemy setup
│   ├── models.py     # Database models
│   └── schemas.py    # Pydantic schemas
├── worker/           # GPU worker
│   ├── demark_world/ # Cloned DeMark-World repo
│   ├── tasks.py      # Celery tasks
│   └── Dockerfile    # GPU Docker image
└── docker-compose.yml

🚀 Quick Start (Local Development)

Prerequisites

• Python 3.12
• Node.js 18+
• Docker & Docker Compose
• FFmpeg

Quick Start

# Run the setup script
./start-dev.sh

# This will start Redis and PostgreSQL
# Follow the printed instructions to start backend and frontend

Manual Setup

1. Clone the Repository

cd aiWatermarkRemover

2. Set Up Environment Variables

Copy .env.template and fill in your values:

cp .env.template .env

You'll need:

• PostgreSQL database URL (or use local Docker)
• Redis URL (or use local Docker)
• Cloudflare R2 credentials (or AWS S3)

3. Start Backend Services (Local)

# Start Redis & PostgreSQL
docker-compose up redis postgres -d

# Install backend dependencies
cd backend
pip install -r requirements.txt

# Run backend
uvicorn main:app --reload

Backend will be available at http://localhost:8000

4. Start Frontend

cd frontend
npm install
npm run dev

Frontend will be available at http://localhost:3000

5. Start Worker (GPU Required)

The worker requires a GPU for processing. For local development without GPU, you can skip this step.

cd worker
pip install -r requirements.txt

# Make sure REDIS_URL and S3 credentials are set
celery -A tasks worker --loglevel=info

📦 Production Deployment

Backend + Database + Redis (Railway)

1. Create a new project on Railway
2. Add PostgreSQL and Redis services from Railway's marketplace
3. Deploy backend:

   cd backend
   railway up
   

4. Set environment variables in Railway dashboard (see .env.template)

Frontend (Vercel)

1. Connect your repo to Vercel
2. Set root directory to frontend/
3. Add environment variable:

   NEXT_PUBLIC_API_URL=https://your-backend.railway.app
   

4. Deploy

Worker (RunPod / Vast.ai)

Since Railway doesn't support GPU instances, deploy the worker separately:

Option 1: RunPod

1. Go to RunPod
2. Create a new pod with A100 40GB or RTX 4090
3. Use the custom Docker image:

   cd worker
   docker build -t your-username/watermark-worker .
   docker push your-username/watermark-worker
   

4. Deploy the image on RunPod
5. Set environment variables (REDIS_URL, S3 credentials)

Option 2: Vast.ai (Cheaper)

1. Go to Vast.ai
2. Select a GPU instance (RTX 3090/4090 recommended)
3. Use the same Docker image as above
4. Configure environment variables

Storage (Cloudflare R2)

1. Create an R2 bucket on Cloudflare
2. Enable public access for the bucket (so users can download results)
3. Get credentials:
   • Account ID
   • Access Key ID
   • Secret Access Key
4. Set in environment:

   S3_ENDPOINT_URL=https://<account-id>.r2.cloudflarestorage.com
   AWS_ACCESS_KEY_ID=<your-key>
   AWS_SECRET_ACCESS_KEY=<your-secret>
   BUCKET_NAME=<your-bucket>
   PUBLIC_URL_BASE=https://pub-xxx.r2.dev
   

🛠️ Technology Stack

• Frontend: Next.js 15, React, Tailwind CSS, TypeScript
• Backend: FastAPI, SQLAlchemy (AsyncPG), Pydantic
• Worker: Celery, PyTorch, DeMark-World
• Database: PostgreSQL
• Queue: Redis
• Storage: Cloudflare R2 (S3-compatible)
• Deployment: Vercel (Frontend), Railway (Backend), RunPod/Vast.ai (Worker)

💰 Estimated Costs (50k minutes/month)

⚖️ Legal Notice

This tool is intended only for removing watermarks from videos that you generated yourself. Using this tool on copyrighted or third-party content may violate laws including:

• DMCA §1202 (US)
• EU Copyright Directive Article 7
• China Copyright Law Article 48

By using this service, you confirm that you have the legal right to remove the watermark from the video.

📝 License

This project uses DeMark-World which is licensed under Apache 2.0, allowing commercial use.

🙏 Acknowledgments

• DeMark-World - The core watermark removal engine
• SoraWatermarkCleaner - The predecessor project

📧 Support

For questions or issues, please open an issue on GitHub.