Onion99/KMP-MineStableDiffusion
Onion99/KMP-MineStableDiffusion has 131 GitHub stars and 11 forks. GitHot measured +1 stars in 24 hours. Last checked 2026-11-16 19:06 UTC. Last push 2026-11-03.
Measured star growth
- in 24h
- +1
- in 7 days
- Collecting history
- in 15 days
- Collecting history
- in 1 month
- Collecting history
Net star changes over overlapping periods, not daily totals. Missing measurements are left blank. Stars measure attention, not software quality.
About
🎨 Multiplatform AI image generation app powered by Stable Diffusion • Built with Kotlin Multiplatform & Compose • Supports SDXL, FLUX, SD3 & more • Native performance via C++/JNI • Android/iOS & Desktop ready
AIREADME.md
OriginalMine StableDiffusion 🎨
The kotlin multiplatform Stable Diffusion client
Generate stunning AI art locally on Your devices

✨ What is Mine StableDiffusion?
Mine StableDiffusion is a native, offline-first AI art generation studio resting right in your pocket or on your desk. Built entirely on modern Kotlin Multiplatform and powered by the blazing-fast stable-diffusion.cpp engine, it delivers desktop-class inference capabilities across Android, iOS, and Desktop platforms.
🎯 The Edge
- 🚀 Native Performance — Pure C++ backend combined with JNI bindings ensures maximum hardware utilization.
- 🔒 100% Privacy — Everything runs offline. No cloud, no subscriptions, no data harvesting.
- 📱 True Multiplatform — A unified, beautiful Compose Multiplatform UX natively adapted for Mobile & Desktop.
- 💠 Pro-Level Controls — Granular control over VRAM (Flash Attention, CPU Offloading, Direct Convolution).
- 🧩 Endless Expansion — Out-of-the-box support for external
.safetensorsLoRAs, advanced Samplers, and metadata injection.
📸 See It In Action
📱 Mobile Experience (Android / iOS)
| App Settings | Generating View | Output Gallery |
|---|---|---|
![]() |
![]() |
![]() |
💻 Desktop Experience (Windows / macOS / Linux)
| Fluid UI | Creative Workflows | macOS Native |
|---|---|---|
![]() |
![]() |
![]() |
🔥 Highlighted Features
🖼️ Batch / Text-to-Image Generation
Never generate just one idea again. Batch Generation synchronously crafts up to 10 iterations of your prompt with beautiful inline progress tracking right in your chat flow.
🎭 Infinite Styles via LoRA
Drag-and-drop support for .safetensors LoRA extensions. Mix multiple LoRAs simultaneously with precision weight sliders built right into the UI.
🎛️ Transparent Sampling & Prompting
Master the machine. Switch Sampler algorithms on the fly (Euler a, DPM++ 2M, LCM, TCD, etc.) to discover exactly how they mutate your art.
🏷️ Self-Documenting Art
Exported PNGs automatically embed all generation parameters (Prompt, Seed, Model, Sampler, CFG, LoRAs). Just drop the image back into any SD tool, and your exact setup is retrieved.
🎲 Model Support & Tiers
We cover bleeding-edge architectures. Ensure the models you pick fit within your device's VRAM limits:
[!TIP] Start Small: We highly recommend testing the waters with SD-Turbo or SD 1.5 models to gauge your device's capabilities before moving to demanding architectures like FLUX.
🎮 Entry & Speed (Fastest, Minimal VRAM)
- ✅ SD-Turbo - Extremely fast 1-step generation
- ✅ SD1.x / SD2.x / Illustrious - The classic standard models
⚖️ Balanced Performance (Moderate VRAM)
- ✅ Illustrious-XL / Pony - Popular high-quality SDXL-based community models
- ✅ SDXL-Turbo - Fast high quality 512x512
- Test:Model

- ✅ SDXL - Standard high-quality base model
- ✅ 🖼️ Z-Image - Advanced image synthesis pipeline
💎 Professional Quality (High Requirements)
Best for high-detail 1024x1024+ generation. Requires modern GPUs with ample VRAM.
- ✅ SD3 / SD3.5 - Stability AI's latest high-fidelity architecture
- ✅ 👁️ Ovis-Image - Vision-language model
- ✅ 🎨 Chroma / Chroma1-Radiance - Vibrant color generation
- ✅ FLUX.1-schnell / dev - Next-gen image quality
- ✅ FLUX.2-dev - Latest and most capable iteration
🛠 Advanced Controls Deep Dive
| Tuning Parameter | What it does | Pro Tip |
|---|---|---|
| Quantization (wtype) | Formats weights to limit RAM footprint (F16, Q8_0, Q4_K). (Note: Not applicable for pre-quantized .gguf files) |
Leave on Auto unless explicitly tuning for low VRAM targets. |
| Offload to CPU | Offloads model computations from GPU to CPU. | Enable if you encounter OOM errors while loading the model. |
| Keep CLIP on CPU | Forces the CLIP text encoder to stay on CPU. | Enable if you experience crashes during image generation. |
| Keep VAE on CPU | Forces the VAE decoder to stay on CPU. | Enable if you encounter OOM errors during decode. |
� Platform Compatibility
| OS | Status | Hardware Requirement |
|---|---|---|
| 🤖 Android | ✅ Stable | Android 11+ (API 30+) running Vulkan 1.2+ |
| 🪟 Windows | ✅ Stable | Windows 10+ running Vulkan 1.2+ |
| 🐧 Linux | ✅ Stable | Standard modern Vulkan 1.2+ drivers |
| 🍎 macOS | ✅ Stable | Silicon or Intel with Metal Support |
| 📱 iOS | ✅ Beta | A13 Bionic or newer, Metal Support |
�️ Architecture
Under the hood, Mine StableDiffusion relies on zero compromises:
graph TB
A[Compose Multiplatform UI] --> B[Cross-Platform ViewModels]
B --> C[Koin Dependency Injection]
C --> D[JNI Bridge]
D --> E[C++ Native Optimization Layer]
E --> F[stable-diffusion.cpp / llama.cpp]
F --> G[Vulkan / Metal Hardware Acceleration]
🚀 Get Started Now
Option 1: Quick Install (Recommended)
- Head over to our Releases Page.
- Download the package crafted for your OS (
.apk,.dmg,.exe). - Simply launch it, point it to a model (
.ggufor.safetensors), and begin typing your prompt!
Option 2: Build From Source
Got Android Studio or IntelliJ IDEA?
# Clone the repository
git clone https://github.com/Onion99/KMP-MineStableDiffusion.git
cd KMP-MineStableDiffusion
# Build for Desktop
./gradlew :composeApp:run
# Build for Android
./gradlew :composeApp:assembleDebug
📚 Resources & Community
- 📑 Read the Changelog — Track what's new.
- 🎨 Community Showcase — Show off your amazing generations or fetch prompts from others!
❤️ Acknowledgements
We stand on the shoulders of giants:
- The masterful leejet/stable-diffusion.cpp
- The trailblazing ggerganov/llama.cpp
- The incredible JetBrains Compose Multiplatform team.
Enjoying the project? Share some love by giving it a ⭐ and spreading the word!
Licensed under the GPL 3.0 License
Links
Linux Do Community: https://linux.do







