🎨 MFLUX-UI - Your Streamlit-Powered FLUX Image Generator!

Transform your wildest ideas into stunning visuals with the power of FLUX models, now with a friendly Streamlit interface!

MFLUX-UI is your friendly neighborhood image generator that wraps the amazing MFLUX project in a beautiful, easy-to-use web interface! Whether you're an AI art enthusiast, a creative professional, or just someone who loves making cool images, MFLUX-UI has got you covered.

• 🖼️ Multiple FLUX Models: Choose from:
  • Schnell - Lightning-fast generation (1-10 steps)
  • Dev - High-quality results with guidance control
  • Kontext - Image-to-image transformations
• 🎛️ Real-time Parameter Tuning: Adjust steps, guidance, dimensions, and more!
• 💾 Smart Output Management: Auto-organized outputs with timestamps
• 🔧 Advanced Options: Quantization, VAE tiling, low-RAM mode
• 📊 Live Generation Progress: Watch your masterpiece come to life!

Before we dive into the fun stuff, make sure you have:

• 🍎 macOS with Apple Silicon (M1/M2/M3/M4)
• 🐍 Python 3.10 or higher
• 💾 At least 16GB RAM (32GB+ recommended for best performance)
• 🗄️ ~50GB free disk space (for model downloads)
• 🤗 Hugging Face account (free!)

git clone https://github.com/jakkaj/mflux-ui.git
cd mflux-ui

We recommend using a virtual environment to keep things tidy:

# Create a virtual environment
python3 -m venv venv

# Activate it
source venv/bin/activate  # On macOS/Linux

The easiest way is using uv:

# Install uv if you haven't already
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install MFLUX
uv tool install --upgrade mflux

Alternatively, you can use pip:

pip install mflux

pip install streamlit

First, create a free account at huggingface.co if you don't have one.

Then, get your access token:

1. Go to huggingface.co/settings/tokens
2. Create a new token (or use an existing one)
3. Copy the token

Set it up in your terminal:

# Option 1: Export as environment variable
export HF_TOKEN="your_token_here"

# Option 2: Use Hugging Face CLI
pip install huggingface-hub
huggingface-cli login
# Paste your token when prompted

Before using FLUX models, you need to accept their licenses on Hugging Face:

• FLUX.1-schnell: Accept here
• FLUX.1-dev: Accept here

Just click "Accept" on each model page while logged into your Hugging Face account.

Fire up the Streamlit interface:

streamlit run runner.py

Your browser should automatically open to http://localhost:8501. If not, just click the link in your terminal!

1. Choose Your Model:

   • Schnell: Super fast, great for prototyping
   • Dev: Best quality, more control
   • Kontext: Upload an image and transform it!

2. Write Your Prompt: Be creative! The more descriptive, the better.

3. Tweak Settings (optional):

   • Adjust steps for speed vs quality
   • Play with guidance scale (Dev/Kontext only)
   • Set custom dimensions
   • Use a fixed seed for reproducible results

4. Click Generate and watch the magic happen!

Your creations are saved in the outputs/ directory with descriptive filenames:

• schnell_20250702_143022.png - Schnell model output
• dev_20250702_143022.png - Dev model output
• kontext_20250702_143022.png - Kontext model output

Make sure you've:

1. Set up your Hugging Face token
2. Accepted the model licenses
3. Have enough disk space for downloads

Try:

• Enabling "Low-RAM mode" in the sidebar
• Using higher quantization (6 or 8)
• Reducing image dimensions
• Closing other applications

The first time you use each model, MFLUX downloads it (~30GB each). This is normal! Subsequent runs will be much faster.

Prompt: "A cozy coffee shop on Mars with Earth visible through the window, digital art style"
Steps: 4
Dimensions: 1024x1024

Prompt: "A majestic dragon made of northern lights dancing across a starry sky"
Steps: 20
Guidance: 3.5
Dimensions: 1024x1024

Upload a photo of your pet and try:

Prompt: "Transform into a renaissance painting with ornate golden frame"
Steps: 20
Guidance: 2.5

• MFLUX Project: github.com/filipstrand/mflux
• FLUX Models: Black Forest Labs
• Hugging Face: huggingface.co
• Streamlit Docs: streamlit.io

Found a bug? Have an idea? Feel free to:

• Open an issue
• Submit a pull request
• Share your coolest generated images!

This project wraps MFLUX, which is based on FLUX models. Please respect all applicable licenses:

• MFLUX is under MIT License
• FLUX models have their own licenses (check Hugging Face pages)
• This wrapper is provided as-is for educational and creative purposes

Now go forth and create some amazing images! Remember, the only limit is your imagination (and maybe your GPU memory 😄).

Made with ❤️ for the AI art community