cog-face-to-many  by fofr

Face stylization via ComfyUI workflows

created 1 year ago
1,344 stars

Top 30.5% on sourcepulse

GitHubView on GitHub
Project Summary

This project provides a ComfyUI workflow to transform any face into various stylized outputs like 3D models, pixel art, claymation, or toys. It's designed for artists and creators looking to leverage AI for rapid character generation and style transfer.

How It Works

The workflow utilizes Stable Diffusion with custom LoRAs and ControlNet models, specifically InstantID, to achieve precise facial style transfer. It allows users to input a reference face and select a desired style, with the model generating a new image matching the style while preserving the identity of the input face.

Quick Start & Requirements

  • Install: Clone the repository, create a Python virtual environment, install dependencies (pip install -r requirements.txt), download specific model checkpoints (.safetensors, .bin), and run ./scripts/clone_plugins.sh to install ComfyUI custom nodes.
  • Prerequisites: Python 3.x, Git, wget, and a GPU with sufficient VRAM for Stable Diffusion models. Specific model files (e.g., albedobaseXL_v13.safetensors, ip-adapter.bin) must be downloaded to designated directories.
  • Running: Execute python3 main.py from the ComfyUI directory.
  • Docs: ComfyUI workflow JSON

Highlighted Details

  • Supports multiple stylistic outputs: 3D, pixel art, claymation, toy, and video game characters.
  • Leverages InstantID for accurate facial identity preservation.
  • Requires specific custom nodes for ComfyUI, including Controlnet Aux, InstantID, IPAdapter Plus, Essentials, and Efficiency Nodes.
  • LoRAs for different styles are sourced from artificialguybr.

Maintenance & Community

  • The project is maintained by fofr.
  • Links to creator support (Patreon, Ko-fi) and Twitter are provided for the LoRA artist.

Licensing & Compatibility

  • The repository itself appears to be under an unspecified license, but the included custom nodes and models may have their own licenses. Users should verify compatibility for commercial use.

Limitations & Caveats

  • Requires manual download and placement of several large model files.
  • Setup involves installing multiple custom nodes, which can be complex for users unfamiliar with ComfyUI's plugin ecosystem.
  • The project relies on external LoRAs and models, whose availability or licensing could change.
Health Check
Last commit

1 year ago

Responsiveness

1 week

Pull Requests (30d)
0
Issues (30d)
0
Star History
20 stars in the last 90 days

Explore Similar Projects

Feedback? Help us improve.