Skip to content

QUICKSTART.md

Latest commit

 

History

History
212 lines (158 loc) · 5.44 KB

QUICKSTART.md

File metadata and controls

212 lines (158 loc) · 5.44 KB

LoraMint Quick Start Guide

Get LoraMint up and running in just a few steps!

🎯 One-Command Start

Option 1: Automatic Setup (Recommended)

The Blazor application will automatically set up and start the Python backend for you.

Linux/macOS:

./start.sh

Windows:

start.bat

That's it! The application will:

  1. 🧹 Stop any existing LoraMint instances (automatic cleanup)
  2. ✅ Check for .NET SDK and Python
  3. 📦 Create Python virtual environment (if needed)
  4. 📥 Install all dependencies (if needed)
  5. 🚀 Start both the Blazor web app and Python backend
  6. 🌐 Open the app at https://localhost:5001

Note: The startup process now includes detailed progress indicators:

  • [STEP 1/3] Virtual environment creation
  • [STEP 2/3] Dependency installation (highlights PyTorch, diffusers, transformers)
  • [STEP 3/3] Backend startup
  • Clear [SUCCESS] and [FAILED] status messages

🔧 Manual Setup (Optional)

If you prefer to set up the Python environment manually first:

Linux/macOS:

./setup-python.sh
./start.sh

Windows:

setup-python.bat
start.bat

📋 What Happens on First Run?

The first time you run LoraMint, it will:

  1. Create Virtual Environment (~30 seconds)

    • Creates an isolated Python environment

  2. Install Dependencies (~2-5 minutes)

    • Installs PyTorch, diffusers, FastAPI, etc.

  3. Setup Wizard (first launch)

    • Detects your GPU and available VRAM
    • Shows model comparison table with compatibility indicators
    • Lets you choose a model (SDXL Base, SDXL Turbo, or Z-Image Turbo)
    • Downloads the selected model (~7-12GB depending on model)

Total first-run time: ~3-5 minutes + model download time

🎨 Using the Application

Once started, open your browser to https://localhost:5001

Complete the Setup Wizard (First Time Only)

  1. Welcome - Introduction to LoraMint
  2. System Check - View your GPU info and available VRAM
  3. Model Selection - Choose from:
    • SDXL Turbo (Recommended) - Fast, 4 steps, good quality
    • SDXL Base 1.0 - Medium speed, 30 steps, high quality
    • Z-Image Turbo - Fast, excellent quality (requires 16GB VRAM)
  4. Download - Wait for model download (progress shown)
  5. Complete - You're ready to generate!

Generate Your First Image

  1. Go to Generate page
  2. Enter a prompt (e.g., "a serene mountain landscape at sunset")
  3. Click Generate Image
  4. Wait ~5-60 seconds depending on model (Turbo models are faster)
  5. View your image!

Train Your First LoRA

  1. Go to Train LoRA page
  2. Enter a name (e.g., "my_style")
  3. Upload 1-5 reference images
  4. Click Start Training
  5. Wait for training to complete
  6. Use your LoRA in future generations!

⚙️ Configuration

Change Model

After initial setup, you can change models via the Settings page:

  1. Click the Settings icon in the navigation
  2. Go to the Models tab
  3. Download additional models if needed
  4. Click Select to switch models or Reload to reload the current model

Disable Auto-Start

Edit src/LoraMint.Web/appsettings.json:

{
  "PythonBackend": {
    "AutoStart": false
  }
}

Then start Python manually:

cd src/python-backend
source venv/bin/activate  # On Windows: venv\Scripts\activate
python main.py

Change Python Backend Path

Edit src/LoraMint.Web/appsettings.json:

{
  "PythonBackend": {
    "Path": "/path/to/your/python-backend"
  }
}

Disable Auto-Install Dependencies

Edit src/LoraMint.Web/appsettings.json:

{
  "PythonBackend": {
    "AutoInstallDependencies": false
  }
}

🐳 Using Docker Instead

If you prefer Docker:

docker-compose up --build

Access the app at http://localhost:5001

🚨 Troubleshooting

Python not found

.NET SDK not found

Port 8000 already in use

  • The startup scripts automatically kill existing processes on port 8000
  • If you still have issues, manually kill the process:
    • Windows: taskkill /F /IM python.exe or check Task Manager
    • Linux/macOS: lsof -ti:8000 | xargs kill -9
  • Or change the port in src/python-backend/main.py: 
    uvicorn.run(app, host="0.0.0.0", port=8001)  # Change to 8001
  • And update appsettings.json: 
    {
  "PythonBackend": {
    "BaseUrl": "http://localhost:8001"
  }
}

GPU not detected

  • Install CUDA toolkit if you have an NVIDIA GPU
  • Otherwise, CPU will be used (much slower)
  • Check GPU status at http://localhost:8000/health

Dependencies fail to install

  • Check your internet connection
  • Try manual setup: ./setup-python.sh or setup-python.bat
  • Check Python version: python --version (should be 3.10+)

📚 Next Steps

  • Read the full README.md
  • Check PROJECT_INSTRUCTIONS.md for details
  • Explore the API docs at http://localhost:8000/docs
  • View Python backend README at src/python-backend/README.md
  • View Blazor web README at src/LoraMint.Web/README.md

🎉 Enjoy!

You're all set to start generating amazing AI images with custom LoRAs!