This guide walks you through setting up the PuLID-Flux integration into ComfyUI for creating stunning results. Follow the steps carefully for a seamless installation.

Installation Steps

Start by cloning the PuLID-Flux repository into your ComfyUI/custom_nodes directory:

git clone https://github.com/sipie800/ComfyUI-PuLID-Flux-Enhanced

If you encounter any issues, use this enhanced version of the repository instead:

git clone https://github.com/al-swaiti/ComfyUI-PuLID-Flux-Enhanced.git

Replace the original repo by removing or renaming it, then use this one.

Install all the dependencies listed in the requirements.txt file into the Python environment where you run ComfyUI.
I recommend avoiding automatic installation scripts, as they can install software without your consent. Stay in control! 😉

pip install -r ComfyUI/custom_nodes/ComfyUI-PuLID-Flux/requirements.txt

Downloading the Models

You will need specific models for the workflow. Place them in the following directories:

• Place the PuLID-Flux pre-trained model in:

  • ComfyUI/models/pulid/

• This model should download automatically during setup and will be saved in the HuggingFace directory.

• If the download fails (causing a face_analysis.py error like init assert 'detection' in self.models), manually download the EVA CLIP model and place it in:

  • ComfyUI/models/clip/

Restart ComfyUI after placing the file.

• Download InsightFace with AntelopeV2, unzip the models, and place them in:

  • ComfyUI/models/insightface/models/antelopev2/

Additional Requirements

The facexlib dependency needs to be installed. Don’t worry—its required models will automatically download upon first use.

OllamaGemini Node

This workflow also uses the ComfyUI-OllamaGemini node. Install it by cloning the repository:

git clone https://github.com/al-swaiti/ComfyUI-OllamaGemini.git

Once everything is set up, you’re ready to use the PuLID-Flux workflow in ComfyUI. Enjoy creating! 🎨