flux controlnet collections

Introduction

The Flux-ControlNet-Collections repository provides a set of ControlNet checkpoints designed for the FLUX.1-dev model by Black Forest Labs, enabling advanced text-to-image generation using Stable Diffusion.

Architecture

This collection supports three models: Canny, HED, and Depth (Midas). Each model is trained on a resolution of 1024x1024 pixels and is optimized for producing highly realistic images. These models are available as version 3 releases, which can be directly utilized in the ComfyUI environment.

Training

The models are trained to operate at a 1024x1024 resolution. The improved v3 versions offer enhanced realism and can be implemented with custom nodes in ComfyUI for efficient workflow management.

Guide: Running Locally

To run the models locally, follow these steps:

1. Clone the Repository: Download the x-flux-comfyui custom nodes from the GitHub repository.
2. Launch ComfyUI: Start the ComfyUI application to interact with the models.
3. Execute Workflows: Use predefined workflows such as canny_workflow.json, depth_workflow.json, or hed_workflow.json for specific models.

Additionally, consider using cloud GPUs for improved performance and scalability, especially for demanding tasks or larger datasets.

License

The models are released under the FLUX.1-dev Non-Commercial License. For more details, refer to the license documentation.