Web UI for interactive, text-guided image generation
Top 82.6% on sourcepulse
Draw2Img provides a user-friendly, web-based interface for interactive image-to-image generation using text prompts. It's designed for users of all skill levels, enabling local, private, and offline creative image manipulation with features like an interactive canvas and real-time generation.
How It Works
The project utilizes a multi-threaded Python websocket server to handle requests and serve a Vue 3 front-end. Image generation is powered by SDXL-Turbo, performing text-guided transformations on a 512x512 canvas. While the server is multi-threaded, a mutex protects the non-thread-safe SDXL-Turbo pipeline, effectively serializing image generation requests. This design prioritizes simplicity and local hosting over high concurrency.
Quick Start & Requirements
pip install .
python draw2img/main.py
Highlighted Details
Maintenance & Community
Contributions are welcome. The project is maintained by GradientSurfer.
Licensing & Compatibility
MIT License for the Draw2Img code. Use of SDXL-Turbo is subject to the Stability AI Non-Commercial License and its acceptable use policy.
Limitations & Caveats
Image generation is effectively single-threaded due to pipeline mutexes, limiting concurrency performance. There is no safety filter for offensive content. For advanced features, projects like Stable Diffusion Web UI or ComfyUI are recommended.
1 year ago
1 day