How to Install Pyramid Flow in ComfyUI:

Getting Pyramid Flow Running in ComfyUI

Here’s the thing—Pyramid Flow doesn’t just work out of the box with a regular ComfyUI install. But it’s not as complicated as it seems. I ran into this recently and figured out a way to set it up without breaking anything else.

First, you’ll need Anaconda. I grabbed it from their site (Grab It From Here) just make sure you skip the PATH environment variable option during install. It’s one of those things that can mess with other setups if you don’t.

Next, I set up a fresh ComfyUI instance. I cloned it straight from GitHub into a new folder (mine was E:\pyramid because I like keeping experiments separate). After that, it was just a matter of running the usual git clone and cd into the directory.

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

For the Python environment, I went with 3.8.10 since it plays nice with the dependencies. Created a new conda env called video, activated it, and installed the requirements. The PyTorch install was straightforward—just copied the command from their site for CUDA 12.1.

conda create -n video python=3.8.10
conda activate video
pip install -r requirements.txt
conda install pytorch==2.2.2 torchvision==0.17.2 torchaudio==2.2.2 pytorch-cuda=12.1 -c pytorch -c nvidia

Once ComfyUI was up (python main.py), I dropped in the PyramidFlow node from AIFSH’s repo. There’s also a speed-optimized node I found useful from ComfyUI-Fast, which helped with generation times.

And that’s it. No magic, just following the steps. If you hit snags, the repos usually have troubleshooting tips. The whole thing took about 20 minutes, mostly waiting for downloads.

Node are here