Abominable Workflows is a collection of ComfyUI workflows that unleashes the full potential of PixArt-Sigma by enhancing its exceptional prompt-following capabilities with the rich detail of SD15 models.
Each workflow generates a distinct style of imagery:

• abominable_PHOTO_ : Realistic images with photographic details.
• abominable_DARKFAN80_ : Dark, cinematic 80s style with VHS aesthetics and dramatic lighting.
• abominable_PIXEL_ : Pixel art style with retro aesthetics and vibrant, blocky shapes.
• abominable_INK_ : Ink-style illustrations with bold outlines and a hand-painted finish.
• abominable_1GIRL_ : Images centered on a captivating woman with photographic realism.
• abominable_MILO_ : European stylized comic book aesthetic with intricate linework.
• classic_abominable_spaghetti : Similar experience to the previous v2 but with all the new enhancements.

Setup Guide

• Required Files
• Required Nodes
• How to Automatically Install the Required Nodes
• Manually Installing Required Nodes on Linux
• Manually Installing Required Nodes on Windows

Required Files

To use the workflows, you need to have the following models installed in ComfyUI.
Place them in the corresponding directories as specified below.

<your_comfyui_dir> / models / checkpoints :

• pixart_sigma-FP16.safetensors (1.2GB)
• photon_refiner-FP16.safetensors (2.1GB)

<your_comfyui_dir> / models / clip :

• t5_xxl_encoder-FP8.safetensors (4.9GB)

<your_comfyui_dir> / models / vae :

• pixart_vae.safetensors (0.1GB)

Required Nodes

[!IMPORTANT] Ensure that your ComfyUI is updated to the latest version.

Additionally, the workflows require the following custom nodes to be installed:

• ComfyUI_ExtraModels: provides support for PixArt-Sigma.
• ComfyUI-Crystools: used for some simple string operations.

How to Automatically Install the Required Nodes

The easiest way to install the required nodes is by using ComfyUI-Manager. This extension for ComfyUI offers management functions to install, remove, disable, and enable custom nodes. It simplifies the process and can save you a lot of trouble.

If for some reason you cannot use ComfyUI-Manager, follow the instructions below to install the nodes manually.

Manually Installing Required Nodes on Linux

To manually install the nodes, open a terminal and run the following commands:

cd <your_comfyui_dir>
git clone https://github.com/city96/ComfyUI_ExtraModels ./custom_nodes/ComfyUI_ExtraModels
git clone https://github.com/crystian/ComfyUI-Crystools ./custom_nodes/ComfyUI-Crystools

If ComfyUI is using a virtual environment, you must activate it before installing the dependencies:

# You might need to replace '.venv' with the directory
# where the virtual environment is located
source .venv/bin/activate

Then, install the dependencies required by the nodes:

python -m pip install -r ./custom_nodes/ComfyUI_ExtraModels/requirements.txt
python -m pip install -r ./custom_nodes/ComfyUI-Crystools/requirements.txt

Manually Installing Required Nodes on Windows

If you are using the standalone ComfyUI release on Windows, open a CMD in the "ComfyUI_windows_portable" folder (the one containing your run_nvidia_gpu.bat file).

From that directory, run the following commands to install the required nodes:

git clone https://github.com/city96/ComfyUI_ExtraModels ComfyUI/custom_nodes/ComfyUI_ExtraModels
git clone https://github.com/crystian/ComfyUI-Crystools ComfyUI/custom_nodes/ComfyUI-Crystools

Then, install the dependencies required by these nodes:

.\python_embeded\python.exe -s -m pip install -r .\ComfyUI\custom_nodes\ComfyUI_ExtraModels\requirements.txt
.\python_embeded\python.exe -s -m pip install -r .\ComfyUI\custom_nodes\ComfyUI-Crystools\requirements.txt

Project Checklist

Workflows

• Editable parameters grouped together.
• Option to select portrait or landscape image orientation.
• Option to choose between two samplers (fast/quality).
• Option for refiner to ignore prompt and focus on other elements.
• Variations by changing the refiner seed.
• Preconfigured workflows with different styles.
• Unified PixArt prompt and refiner prompt.
• CivitAI prompt extractor compatibility.
• Support for PixArt 2K model.
• Automatic detection of the model used and the optimal image size.

Development & documentation

• Explanation of how to install required files.
• Explanation of how to manually install nodes on Linux and Windows.
• Models in FP16 or FP8 for minimal VRAM (6 or 8GB).
• Apply ComfyUI advances to speed up inference (own nodes?).
• Simplify installation with minimal custom nodes.

Lincense

Copyright (c) 2024 Martin Rizzo
This project is licensed under the MIT license.
See the "LICENSE" file for details.

Acknowledgments

I would like to thank the developers of PixArt-Σ for their outstanding work. Their model has been a crucial component in the development of my workflows.

Further Information about PixArt-Σ:

• PixArt-Σ GitHub Repository
• PixArt-Σ Hugging Face Model
• PixArt-Σ arXiv Report