Lumina AI Photo Enhancer

An enterprise-grade, full-stack AI photo enhancement web application orchestrating LLaMA-3, BLIP, and ESRGAN models.

Key Features • Tech Stack • Installation • Evaluation Framework

Overview

Lumina Web is a production-ready application that allows users to seamlessly upload, automatically categorize, and geometrically enhance their photographs using state-of-the-art AI. The system dynamically routes images through a network of Replicate models to upscale, denoise, color grade, or fix low-light photography.

The application utilizes a sleek, responsive, glass-morphic Tailwind CSS frontend heavily inspired by modern macOS design paradigms.

Key Features

• 🔐 Secure OAuth Authentication: Google OAuth via Passport.js & PostgreSQL.
• ✨ AI Suite: Real-ESRGAN (Upscaling), CodeFormer (Restoration), and Night-Enhancement.
• 🧠 Smart Albums: Salesforce BLIP & Meta LLaMA-3 intelligently auto-tag and group images.
• 💳 Subscription Gateway: Interactive pricing modal mapped to user limits.
• ☁️ Permanent Cloud Storage: Images proxied securely to Cloudinary with automatic watermarking.
• 📊 Admin Dashboard: Secure /admin.html tracking real-time KPIs and system telemetry.

Tech Stack

• Frontend: Vanilla HTML5, JavaScript (ES6+), Tailwind CSS, PostCSS
• Backend: Node.js, Express.js (REST API)
• Database: PostgreSQL (via pg pool), connect-pg-simple (Session Store)
• Authentication: Passport.js (Google Strategy)
• Cloud Storage: Cloudinary API
• AI Provider: Replicate API
• Data Science (IQA): Python 3, OpenCV, pandas, scikit-image, ReportLab

---

Installation & Setup

To clone and run this application locally, you'll need Git and Node.js (which comes with npm) installed on your computer.

1. Clone the repository

git clone https://github.com/yourusername/Lumina.git
cd Lumina

2. Install Dependencies

npm install

3. PostgreSQL Database Setup

Ensure you have a local or cloud instance of PostgreSQL running. Create a new database named lumina.

Run the automated setup scripts to inject the secure UUID-based schema:

node db/setup.js
node db/migrate_eval.js

4. Environment Variables

Create a .env file in the root directory and populate it with your confidential API Keys:

# Server
PORT=3000
SESSION_SECRET=your_super_secret_string

# PostgreSQL Database
DATABASE_URL=postgresql://postgres:YOUR_PASSWORD@localhost:5432/lumina

# OAuth Integration
GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret
OAUTH_CALLBACK_URL=http://localhost:3000/auth/google/callback

# APIs
REPLICATE_API_TOKEN=your_replicate_token
CLOUDINARY_URL=cloudinary://YOUR_KEY:YOUR_SECRET@YOUR_CLOUD_NAME

5. Start the Application

# Starts the server via nodemon on localhost:3000
npm start

---

Evaluation Framework

Lumina features a robust telemetry and Image Quality Assessment (IQA) backend.

The backend inherently measures the latency (processing_time_ms) of every Replicate API call. Additionally, users can leave subjective feedback via a Thumbs Up/Down widget on the Frontend slider.

Objective Quality Mathematical Scoring

You can mathematically score the AI generations against geometric and spatial distortion properties. Ensure you have installed the Python dependencies:

pip install -r scripts/requirements.txt

Run the Evaluator:

python scripts/eval_quality.py

Compile PDF Report:

python scripts/generate_report.py

---

Note: For security reasons, the /admin.html dashboard natively redirects users away unless their PostgreSQL users row has the is_admin boolean precisely set to TRUE.