Setup ComfyUI from Scratch to Run Flux diffusion model on RunPod – Step by Step Tutorial!

Welcome to our comprehensive, step-by-step tutorial on setting up ComfyUI to run the Flux Schnell diffusion model on RunPod! In this video, we’ll walk you through everything you need to get started—from creating a RunPod network disc and configuring a new pod, to installing ComfyUI and the ComfyUI Manager from scratch, and finally downloading the required Flux Schnell model files from HuggingFace. By the end of this tutorial, you’ll have a fully functional image-generation workflow using the Flux Schnell model!

Video

Commands

Here are the commands which are shown in the video.

apt update

mkdir /content
cd /content
git clone https://github.com/comfyanonymous/ComfyUI.git
cd ComfyUI

python -m venv venv
source venv/bin/activate

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
pip install -U --pre xformers
pip install -r requirements.txt

cd /content/ComfyUI/custom_nodes
git clone https://github.com/ltdrdata/ComfyUI-Manager comfyui-manager

# Download Models

mkdir -p /workspace/ComfyUI/models/checkpoints
# In the video I forgot to cd to checkpoints. 
# I have adjusted the commands here so that you can just refer to these commands later
# instead of going through the whole video again and again. 
cd /workspace/ComfyUI/models/checkpoints
wget https://huggingface.co/Comfy-Org/flux1-schnell/resolve/main/flux1-schnell-fp8.safetensors

mkdir -p /workspace/ComfyUI/models/clip
cd /workspace/ComfyUI/models/clip
wget https://huggingface.co/comfyanonymous/flux_text_encoders/resolve/main/clip_l.safetensors
wget https://huggingface.co/comfyanonymous/flux_text_encoders/resolve/main/t5xxl_fp16.safetensors

mkdir -p /workspace/ComfyUI/models/vae
cd /workspace/ComfyUI/models/vae
wget https://huggingface.co/black-forest-labs/FLUX.1-schnell/resolve/main/ae.safetensors

mkdir -p /workspace/ComfyUI/models/controlnet
cd /workspace/ComfyUI/models/controlnet
wget https://huggingface.co/InstantX/FLUX.1-dev-Controlnet-Union/resolve/main/diffusion_pytorch_model.safetensors

find /workspace/ComfyUI -type f -exec ls -lh {} \;

apt update
apt install rsync

rsync -ah --progress  /workspace/ComfyUI/models/controlnet/ /content/ComfyUI/models/controlnet/
rsync -ah --progress /workspace/ComfyUI/models/vae/ /content/ComfyUI/models/vae/
rsync -ah --progress /workspace/ComfyUI/models/clip/ /content/ComfyUI/models/clip/
rsync -ah --progress /workspace/ComfyUI/models/checkpoints/ /content/ComfyUI/models/checkpoints/

cd /content/ComfyUI/
# activate venv if not done already: source venv/bin/activate
python main.py --listen 0.0.0.0 --port 4000Code language: PHP (php)

Download the flux schnell workflow from here

Source Link: Setup ComfyUI from Scratch to Run Flux diffusion model on RunPod – Step by Step Tutorial!

0
Subscribe to my newsletter

Read articles from Mobisoft Infotech directly inside your inbox. Subscribe to the newsletter, and don't miss out.

Written by

Mobisoft Infotech
Mobisoft Infotech

Mobisoft Infotech is a software product engineering company dedicated to providing value, quality, and lasting business expansion through digital innovation, change, and the adoption of technology. Our proficiency in digital technology encompasses Mobile, Cloud, DevOps, Web, IoT, AI, UI/UX, Testing, Robotic Process Automation, and various other digital transformation services, catering to companies across a wide range of industries. With more than ten years of experience in constructing digital solutions, offering consulting, and handling outsourcing, Mobisoft has assisted clients from over 30 countries in embracing digital transformation, capitalizing on market prospects, and attaining sustained outcomes.