tianjiangqiji/nova-image-studio
自托管的 AI 图像生成工作台 · 自定义模型 · 多模式 · PWA · 实时任务 支持Agent模式，UI设计模式，工作台模式，无限画布，反推提示词，提示词广场，GIF生成。前后端任务机制轻量后端；三端兼容 UI：桌面端、平板端、移动端自适应布局
What it solves
Nova Image Studio is a self-hosted AI image generation workbench designed for individuals and teams. It provides a unified interface to manage multiple AI models, handle image generation tasks asynchronously, and offer specialized workflows for UI design and agent-based creation, removing the need to switch between different AI service providers.
How it works
The project uses a Next.js 16 and React 19 frontend (exported as a PWA) and a lightweight Node.js backend. The backend manages a task queue using SQLite and WebSockets for real-time status updates. It acts as a proxy to various image and text AI APIs (such as OpenAI and Google Gemini), allowing users to configure their own API keys and base URLs directly in the browser's local storage.
Who it’s for
It is intended for designers, developers, and AI enthusiasts who want a centralized, self-hosted hub for image generation, UI prototyping, and AI-assisted asset creation.
Highlights
- Six Specialized Modes: Includes Text-to-Image, Image-to-Image, an Agent-based chat workspace, a UI Design mode, reverse prompting, and GIF generation.
- UI Design Workflow: A comprehensive pipeline that converts UI screenshots into sliced assets and then uses an AI agent to recreate the design as actual HTML/CSS/JS code.
- Customizable Model System: Supports independent configuration for image and text models across different protocols (OpenAI, Gemini, Anthropic).
- Real-time Task Management: Implements a concurrent task queue with WebSocket notifications and automatic cleanup of expired results.
- PWA Support: Can be installed as a desktop app with a responsive UI that adapts to desktop, tablet, and mobile screens.
Related
- Project
- Project
- Project
- Project
- Project