This repository provides a Next.js 14 and Tailwind CSS 3 template for an AI image outpainting website, targeting users who want to build a functional AI art generation platform. It offers a pre-built frontend with internationalization, payment integration, and Google login, aiming to simplify the setup of such services.

How It Works

The frontend communicates with a backend via a Message Queue (MQ) system. Upon order creation, an SSE request is sent to an MQ queue manager (a Python microservice). GPU servers listen to this queue, process AI outpainting jobs using models like those from Hugging Face, upload results to Cloudflare R2, and notify the website via callbacks. This decoupled architecture allows independent scaling of GPU resources without affecting the website. An alternative polling-based approach is also described.

Quick Start & Requirements

• Install dependencies: yarn install
• Migrate database: yarn pg:migrate
• Run development server: yarn dev
• Prerequisites: Node.js, Python (for MQ manager), PostgreSQL database, Cloudflare R2 storage, Google API credentials, PayPal/Stripe API keys.
• Internationalization: Requires yarn extract and yarn translate (with OpenAI API key).
• Demo: https://www.ai-outpainting.com/

Highlighted Details

• Next.js 14, Tailwind CSS 3, NextUI, React Icons.
• LinguiJS for internationalization, with automated translation scripts.
• Integrated Google login via NextAuth.js.
• Supports PayPal and Stripe payments.
• Decoupled backend architecture using MQ and R2 storage for scalability.
• Includes scripts for generating website content and managing translations.

Maintenance & Community

• Project is actively maintained with recent updates addressing build issues and Google login configuration.
• Contact for technical discussion: WeChat (fafafa-ai).
• Accepts paid external link submissions.

Licensing & Compatibility

• MIT License.
• Requires attribution link to https://www.ai-outpainting.com/.
• Compatible with commercial use under MIT terms.

Limitations & Caveats

Local Google login setup requires specific code modifications to resolve issuer mismatches. The backend AI model and MQ manager are not included and must be self-hosted or integrated via services like Runpod. Translation scripts rely on external APIs and require configuration.