I have been quietly building a small Windows app called Gentastic, and it is now open source. It generates images locally, on your own GPU.
The reason it exists is simple. Most local image tools make you learn something before you can make a single picture: ComfyUI hands you a node graph to wire up, and the friendlier apps are not always stable. Gentastic is aimed at people who find ComfyUI too hard and who are tired of running into bugs in tools like Amuse AI, and the whole goal is to stay simple enough for non-technical users. You pick a model, type a prompt, and press Generate.
Everything runs on your machine. No cloud, no account, no subscription, no Python, no telemetry, and nothing you type or generate leaves your computer.
What it does today:
- Text-to-image and image-to-image
- Instruction-based editing with FLUX.1 Kontext and FLUX.2 klein ("put them on a beach", "change the suit to grey")
- Keep-face identity from a single reference photo, using PhotoMaker
- One-click model downloads from Hugging Face, plus presets and a gallery that stores each image's settings inside the PNG
- Sane defaults and a short explanation on every field, so you can start before you know what CFG or a sampler is
- Automatic CUDA, Vulkan or CPU detection, with an optional on-demand CUDA runtime so you do not need the CUDA Toolkit installed
Models: FLUX.2 klein (the fast default, around 4 steps), FLUX.1 schnell and dev, and the SDXL family (Illustrious, RealVisXL, Pony, NoobAI). The FLUX models ship as quantized GGUF to keep downloads small.
A note on responsible use. Gentastic can reproduce a real person's face and edit real photographs. Running offline keeps your work private, but it does not make you any less responsible for what you create. Please only use images you have the right to use, respect people's consent and likeness, and stay within the law where you live. The identity and adult features are hidden by default and ask you to confirm before you enable them.
It is free and released under a BSD 3-Clause license. It is also a one-person project, so if you find it useful, a GitHub star, a bug report, or a coffee genuinely helps me keep it going.
Website: https://idct.tech/gentastic
Source and issues: https://github.com/ideaconnect/gentastic