Universal native Generative AI Studio for Omarchy Linux and Quickshell (Wayland / Hyprland).
Designed for high-performance local diffusion inference with zero cloud dependencies and zero token costs.
QIS connects the reactive QtQuick/QML interface with native system acceleration and local AI engines:
graph LR
subgraph UI ["Omarchy Desktop (Wayland / Hyprland)"]
Bar["Quickshell Bar Widget"]
Panel["QIS Studio Panel (QML / #screens)"]
end
subgraph Core ["Native Performance Layer"]
Bridge["Rust Bridge (bin/qwen-bridge)"]
CLI["Lifecycle Manager (qis CLI)"]
end
subgraph Backend ["Local AI Engines (NVIDIA CUDA / ROCm)"]
Comfy["ComfyUI Backend (DiT Q4_K_M GGUF)"]
Ollama["Ollama / Vision (Prompt Optimizer & WD14)"]
end
Bar <--> Panel
Panel <--> Bridge
CLI <--> Bridge
Bridge <--> Comfy
Bridge <--> Ollama
| Studio & Canvas | Prompt & Inspiration Explorer | History & Local Archive |
|---|---|---|
![]() |
![]() |
![]() |
- Text-to-Image (DiT Diffusion): Native high-resolution generation via local ComfyUI backend.
- Image-to-Image & Auto Aspect Match: Automatic aspect ratio extraction (
21:9,16:9,4:3,1:1,3:4,9:16) preventing image distortion. - WD14 Tag Interrogator: Fast Danbooru tag extraction via ONNX / ViT models.
- Vision Prompt Engineering: Reverse-engineering existing scenes into rich diffusion prompts.
- Civitai Explorer: Curated prompts, negative prompts, and generation settings browser.
- Fast Rust Bridge: High-performance native bridge (
bin/qwen-bridge) executing telemetry, PNG metadata chunk manipulation, and system monitoring in sub-millisecond speeds. - Live System Telemetry: Header badges showing connected Qwen model, ComfyUI, Ollama, CUDA, Quickshell, and VRAM utilization.
- In-Button Live Timers: Dynamic, non-intrusive second counters inside active buttons (
GENERATE,UPDATE,OPTIMIZE,TAGS). - Wallpaper Integration: One-click wallpaper deployment to Omarchy and Hyprland.
| Action / Gesture | Trigger | Result |
|---|---|---|
| Toggle Studio Panel | Click on bar widget or omarchy-shell simonez.qwenimage toggle |
Opens/closes the floating studio window |
| Quick Plugin Update | Left Click on UPDATE in header |
Updates plugin & Rust bridge without restarting the shell (< 1s) |
| Full Stack Update | Right Click on UPDATE in header |
Opens confirmation banner to update ComfyUI core, custom nodes, and models |
| Close Studio | Esc or click outside |
Smoothly closes the window and frees viewport |
| Live Telemetry Refresh | Click on REFRESH |
Forces immediate hardware and VRAM telemetry update |
QIS automatically profiles your hardware on installation and suggests the optimal runtime configuration:
| Tier | VRAM | Recommended Configuration |
|---|---|---|
| Tier S |
|
Full FP8 / BF16 direct weights, maximum generation speed. |
| Tier A | 8 – 12 GB | GGUF Q4_K_M DiT + Qwen3-VL 8B Text Encoder with asynchronous VRAM offloading (--async-offload 2). |
| Tier B | 6 GB | GGUF Q3_K_S DiT + Qwen2.5-VL 3B Text Encoder with sequential CPU offload. |
| Tier C |
|
Remote LAN cluster execution via network ComfyUI backend. |
git clone https://github.com/simonezpx3/omarchy-qwenimage.git
cd omarchy-qwenimage
chmod +x install.sh
./install.shThe installer will:
- Profile GPU vendor (NVIDIA, AMD, Intel) and detect available VRAM.
- Check for ComfyUI and diffusion models; if missing on new hardware, offer one-click automated stack installation.
- Deploy the native bridge binary
bin/qwen-bridge. - Deploy the
qismanagement CLI into~/.local/bin/qis. - Validate plugin schema via
omarchy plugin validate. - Register the plugin cleanly into the Omarchy bar layout.
- Perform safe IPC hot-reload.
Once installed, manage the entire ecosystem with the qis command:
qis # Interactive management menu (Setup, Update, Models, Rollback)
qis status # Displays dependency health matrix and hardware profile
qis setup # Automated installation of ComfyUI, custom nodes, and models
qis update # Check and update QIS plugin, ComfyUI core, and nodes
qis rollback # Revert ComfyUI to the previous working checkpointThis plugin strictly complies with the Omarchy Marketplace Security Baseline:
- Pinned Git Commits: ComfyUI core and all custom nodes are pinned to immutable commit SHAs.
- Cryptographic Hash Verification: All Python packages and sub-dependencies are locked with strict SHA-256 hashes (
pip install --require-hashes) across CUDA 12.8, ROCm 6.2, and CPU architectures inlocks/. - Zero Unpinned Execution: No unhashed or arbitrary code is downloaded during clone, install, or update operations.
cd omarchy-qwenimage
chmod +x uninstall.sh
./uninstall.shMIT License. Developed by simonez.


