Flux Installation Guide for LightDiffusion-Next¶
Welcome to the Flux installation guide for LightDiffusion-Next. Follow the steps below to set up Flux on your system.
Table of Contents¶
Introduction¶
Flux is a powerful AI model compatible with LightDiffusion-Next that offers enhanced image generation capabilities. This guide will walk you through the process of setting up Flux to work with LightDiffusion-Next.
Prerequisites¶
Before you begin, ensure you have the following:
- At least 25GB of free space on your hard drive (40-50GB recommended)
- A CUDA-compatible GPU with at least 6GB VRAM (16GB+ recommended)
Run the Application¶
Windows¶
Open a command prompt and execute the run.bat
file to start the application:
./run.bat
Linux¶
Open a terminal and run the run.sh
script to launch the application:
chmod +x run.sh
./run.sh
Using Flux¶
To use Flux with LightDiffusion-Next:
Via GUI¶
- Launch LightDiffusion-Next.
- Select the Flux model from the dropdown menu.
- Configure your generation parameters.
- Click “Generate” to create images using Flux.
Via CLI¶
Use the following command to generate images with Flux:
./pipeline.bat "your prompt here" width height number_of_images batch_size --flux
For example:
./pipeline.bat "A beautiful sunset over the ocean" 1024 768 1 1 --flux
Tips and Tricks¶
- Recommended Resolutions: Flux works best with specific resolutions. Refer to this guide for optimal resolution choices.
- Memory Management: Flux requires significant VRAM. Adjust batch sizes and resolution settings to optimize memory usage.
- Experiment with Prompts: Flux may respond differently to prompts compared to standard models. Experiment with prompt styles to find what works best.
Troubleshooting¶
If you encounter any issues while using Flux, consider the following:
- Check VRAM: Ensure you have enough VRAM available. Flux models typically require more memory than standard models.
- Update Drivers: Make sure your GPU drivers are up to date to avoid compatibility issues.
- Reduce Resolution: If you’re running out of memory, try reducing the resolution or batch size.
For additional help, please refer to the GitHub issues page or join the community discussion forum.
Wish you good generations!