techjarves/Local-AI-Image-Generator — reverse-engineered prompt

Build me a simple Windows app for running Stable Diffusion locally without making the user install Python or mess with command lines. I want someone to be able to unzip the folder, double click a start file, and get a local browser workspace that just works.

The app should set up its own portable runtime inside the project folder, detect the person’s GPU, use CUDA for Nvidia, Vulkan for AMD or Intel Arc, and fall back to CPU if needed. It should support Safetensors, GGUF, and CKPT models. Users should be able to drop model files into a models folder, drag and drop them in the interface, or paste a Hugging Face link to download one.

Make the interface feel like a small image generation studio with a prompt box, model picker, basic generation settings, live CPU, RAM, GPU, and VRAM stats, and a gallery of saved PNG images with their prompt settings saved too. Keep everything offline and private once models are downloaded.