tianjiangqiji/nova-image-studio

What it solves

Nova Studio is a self-hosted AI generation workbench designed for individuals and small teams to manage image and video creation across multiple AI models. It eliminates the need to switch between different AI service providers by providing a unified interface for text-to-image, image-to-image, video generation, and AI-assisted UI design.

How it works

The project uses a Next.js and React frontend (as a PWA) and a Node.js backend with SQLite and WebSockets for task scheduling. It is designed to be model-agnostic: users configure their own API keys and base URLs for various text and image models.

For video generation, it uses a plugin-based architecture where the host provides the infrastructure (queue, history, media upload), and JSON-based plugins define the specific API protocols, UI forms, and polling logic for different video providers.

Who it’s for

• AI Content Creators: Users who want a single dashboard to access multiple image and video models.
• UI/UX Designers: Those looking to convert UI mockups into sliced assets and functional web pages using AI agents.
• Developers: People who want a self-hosted, extensible workbench with a plugin system for adding new AI capabilities.

Highlights

• Seven Working Modes: Includes text-to-image, image-to-image, a conversational AI agent with browser CDP tools, GIF generation, and a reverse prompt tool.
• Plugin-Driven Video Workbench: Allows adding new video generation capabilities via pure JSON plugins without changing the core code.
• UI Design Mode: Features AI-powered image slicing, background filling, and a multi-turn agent that reproduces mockups as HTML/CSS/JS.
• Agentic Capabilities: The AI agent can perform web searches, reason, and interact with a local Chrome browser to extract data from pages like Taobao.
• Flexible Model System: Supports multiple protocols including OpenAI, Gemini, and Anthropic, with per-model configuration stored in the browser's localStorage.