This is the C# / .NET version of my old [Flux Image Generator](https://civitai.com/articles/6754/flux-image-generator) — a Windows desktop app that drives Replicate with the Flux 2 family, GPT-Image (1.5 and 2), and Google's nano-banana-2, with an in-app gallery and the full prompt-and-parameters recipe baked into every saved image.
Update 04.05.2026 — v1.0.0 shipped. In-app gallery, concurrent queue, persisted prompt and model, EXIF/PNG-text recipes, drag-and-drop input images. Single-file Windows exe, no installer, no .NET runtime to install separately.
Update 04.05.2026 — The catalogue now extends well beyond Flux: `openai/gpt-image-1.5`, `openai/gpt-image-2` and `google/nano-banana-2` are wired in alongside the Flux 1.1 and Flux 2 line-ups.
Update 08.03.2025 — Back on the old article I said "I am currently working on a C#, .NET version." That C#, .NET version is this one. 🙂
Initial layout
Result card
Gallery
Program Description: Image Generator MAUI
A proper Windows desktop app — not a browser tab tied to a running Python process — built on .NET MAUI. It drives Replicate, so any model the API exposes is reachable. Out of the box you get Flux 1.1 Pro and Pro Ultra, the Flux 2 family (Klein 4B / Flex / Pro / Max), OpenAI's gpt-image-1.5 and gpt-image-2** (hosted on Replicate), and Google's nano-banana-2 with its 15-value aspect picker and 1K / 2K / 4K resolution selector. Click Refresh Models and the catalogue re-fetches — new models surface without rebuilding the app.
Features
- Multi-model catalogue: refreshable on demand from Replicate
- Concurrent queue: fire a second prompt before the first finishes — each click snapshots its own in-flight job
- Persisted state: the last prompt and model survive a restart
- In-app gallery: tile grid sorted by date / name / size, live updates as new images land
- Recipe baked in: every saved image carries the prompt, seed and model knobs in EXIF (JPG / WebP) or the PNG `Comment` chunk — paste it back into the gallery, MediaInfo or `exiftool` months later and reproduce
- Drag-and-drop: pull input images straight from Explorer onto the prompt panel
- Saved here: `%USERPROFILE%\Pictures\ImageGeneratorMAUI\`
Why?
The Gradio frontend was fine for a single Flux model, but Replicate's catalogue has grown a lot since 2024 and a one-tab-per-launch Python app could not keep up. I wanted something I open like any other program, that remembers where I left off, and that lets me browse what I have already made without leaving it. So I rebuilt it.
Download
GitHub release (https://github.com/MR-444/ImageGenerator.MAUI/releases) — single-file Windows exe, no installer. On first run Windows SmartScreen will warn you because the exe is not code-signed (certificates are expensive for a hobby project); click **More info → Run anyway**.
You will still need a [Replicate API token](https://replicate.com/account/api-tokens) and a bit of credit on your account 😉 — most images come in around $0.003–$0.05.
Source is MIT-licensed: <https://github.com/MR-444/ImageGenerator.MAUI>. Bug reports and feature requests are welcome on the issue tracker.
If you give it a try, drop a comment with what you generated — I read every one.