PhotoGen

Your passport photo, done right. For free.

AI-powered face detection, background removal, and auto-cropping
for 10+ countries. No signup. No watermarks. No data stored.

Demo · Features · Getting Started · Supported Countries · Tech Stack · API · Deployment · License

Demo

photogen.mov

✨ Features

AI Processing Face Detection via MediaPipe BlazeFace with Haar cascade fallback Background Removal using BiRefNet-portrait + alpha matting Auto-Crop to exact government specs (head %, eye position, dimensions) Enhancement with white balance + adaptive CLAHE contrast	User Tools Post-Processing sliders for brightness, contrast, saturation Manual Crop with oval face template and alignment guides Before / After side-by-side comparison slider Print Sheet 4-up grid ready for photo center printing
Output Quality 300 DPI publication-quality JPEG export Compliance Validation for dimensions, head size, eye position, file size 10+ Country Specs including US, EU, UK, Canada, Australia, India, China, Japan	Zero Friction Wide Format Support for HEIC, AVIF, JPEG, PNG, WebP, BMP 100% Free with no accounts or watermarks Privacy First with no data stored on server

🔄 Processing Pipeline

graph LR
    A[📷 Upload] --> B[🎯 Face Detection]
    B --> C[✂️ Background Removal]
    C --> D[📐 Auto-Crop]
    D --> E[🎨 Enhancement]
    E --> F[💎 300 DPI Output]

Loading

🚀 Getting Started

Prerequisites

Requirement	Version
Python	3.11+
Node.js	18+

Backend

cd backend
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
uvicorn app.main:app --reload --port 8000

The ML model (~170MB BiRefNet) downloads automatically on first request.

	URL
API	http://localhost:8000
Docs	http://localhost:8000/docs

Frontend

cd frontend
npm install
npm run dev

Open http://localhost:3000

🌍 Supported Documents

Document	Country	Size	Head Height	Background
US Passport	🇺🇸 United States	2x2" (600x600px)	50-69%	White
US Visa	🇺🇸 United States	2x2" (600x600px)	50-69%	White
EU Passport	🇪🇺 EU / Schengen	35x45mm (413x531px)	70-80%	Light Gray
UK Passport	🇬🇧 United Kingdom	35x45mm (413x531px)	70-80%	Light Gray
Canada Passport	🇨🇦 Canada	35x45mm (420x540px)	71%	White
Australia Passport	🇦🇺 Australia	35x45mm (413x531px)	75%	White
India Passport	🇮🇳 India	35x45mm (413x531px)	75%	White
China Visa	🇨🇳 China	33x48mm (390x567px)	70%	White
Japan Passport	🇯🇵 Japan	35x45mm (413x531px)	70%	White
Germany Passport	🇩🇪 Germany	35x45mm (413x531px)	75%	Light Gray

🛠 Tech Stack

Backend	Frontend
Library Purpose FastAPI + Uvicorn API server MediaPipe Face detection OpenCV Haar cascade fallback rembg (BiRefNet) Background removal PyMatting Alpha matting Pillow + pillow-heif Image I/O scikit-image CLAHE enhancement	Library Purpose Next.js 16 + React 19 App framework TypeScript Type safety Tailwind CSS 4 Styling react-easy-crop Manual crop Axios HTTP client

📡 API

Method	Endpoint	Description
POST	/api/process	Full processing pipeline
POST	/api/detect-face	Face detection only
POST	/api/remove-background	Background removal only
GET	/api/requirements	All country photo specs
GET	/api/requirements/{code}	Single country spec
GET	/api/health	Health check

📁 Project Structure

Click to expand

photogen/
├── backend/
│   ├── app/
│   │   ├── api/routes/
│   │   ├── core/
│   │   ├── services/
│   │   ├── models/
│   │   └── data/
│   ├── Dockerfile
│   └── requirements.txt
├── frontend/
│   ├── app/
│   └── src/
│       ├── components/
│       ├── hooks/
│       ├── lib/
│       └── constants/
└── shared/
    └── photo_requirements.json

🚢 Deployment

Service	Platform	Notes
Frontend	Vercel	Edge CDN, auto-deploy from main
Backend	HuggingFace Spaces	Docker, 16GB RAM, 2 vCPU

The backend Dockerfile is configured for HuggingFace Spaces with pre-downloaded models.

📄 License

MIT