Description
nvzhuang is a GitHub repository offering a virtual try-on website that leverages AI to generate female clothing reference images. The project includes a collection of 1407 AI-generated reference images and supports both English and Chinese search queries, making it accessible to a broad user base. It integrates with popular AI image generation services such as OpenAI Images, Google Gemini, and Doubao, and provides flexibility for users to configure custom Base URLs, API Keys, and models.
Read More
A core focus of the nvzhuang project is security and robust deployment. The `AGENTS.md` document outlines strict security guidelines for AI agents and collaborators, emphasizing a "zero leakage principle." This includes absolute prohibition of committing real API keys, private tokens, passwords, or certificate keys to the Git repository. Instead, only template configuration files with empty or placeholder key fields are allowed. Personal privacy and sensitive information are also strictly forbidden in code, comments, commit messages, or PR descriptions.
The backend service, built with GoFrame, incorporates defensive coding practices. It ensures multi-layer fallback mechanisms, such as automatically loading `config.example.yaml` if `config.yaml` is not explicitly created. The service port is hardcoded to `8000` when not configured, preventing random high-port assignments. Dockerfile and Docker Compose configurations are designed for out-of-the-box readiness, ensuring template configurations are in place during build and orchestration phases.
Frontend user configurations, including personal model settings and historical generation records, are stored locally in the user's browser (localStorage/IndexedDB) and are not persisted to disk or server-side logs by the backend. The project also implements strict validation for external inputs and URL parameters, allowing only same-origin or local loopback (`127.0.0.1`/`localhost`) for sensitive parameters like `?tryon=`. This prevents silent hijacking of requests and exfiltration of private API keys or user photos to untrusted third-party domains. Public remote addresses are strictly enforced to use HTTPS for credential transmission.
Deployment instructions are provided for Docker Compose, Docker Run, and direct command-line execution, all highlighting the built-in fallback to port 8000. The frontend prioritizes backend addresses from user settings, global variables, and securely validated URL parameters, with a local fallback to `http://127.0.0.1:8000` for local access. The project aims to provide a secure, flexible, and easy-to-deploy solution for AI-powered virtual try-on experiences.
- CategoryAI Product Photography
- Pricing ModelOpen source
- DeploymentCloud, Self-hosted
- PlatformsWeb
nvzhuang's Core Features
1407 AI-generated female clothing reference images
Virtual try-on website functionality
Supports English and Chinese search
Integration with OpenAI Images
Integration with Google Gemini
Integration with Doubao (豆包)
Customizable Base URL, API Key, and AI model
Strict security guidelines for AI agents and collaborators
Prohibition of committing real API keys to Git repository
Strict validation for external URL parameters (e.g., `?tryon=`)
Automatic fallback to `config.example.yaml` if `config.yaml` is missing
Default backend service port set to 8000
Frontend user configurations stored locally in browser
HTTPS enforcement for public remote backend addresses
Docker Compose and Docker Run deployment options
Getting Started with nvzhuang
Clone the repository: Obtain the project files from the GitHub repository.
Configure environment variables: Copy `backend/.env.example` to `backend/.env` and fill in your API keys (e.g., `ARK_API_KEY`, `OPENAI_API_KEY`, `GEMINI_API_KEY`).
Deploy with Docker Compose: Run `docker compose up -d` to build and start the backend container, which will listen on port 8000 by default.
Deploy with Docker Run (alternative): Use `docker run` command with appropriate environment variables and port mapping.
Run directly from source (alternative): Navigate to the `backend` directory, set environment variables, and execute `go run main.go`.
Access the frontend: The frontend static page will automatically connect to the backend, prioritizing user settings, global variables, or local fallback.
nvzhuang's Use Cases
- Virtual Clothing Try-On
- AI Image Generation Exploration
- Fashion Design Inspiration
- Secure AI Application Development
- Open-Source AI Project Deployment
