Hugging Face's logo

bbbboiwow
/
cocccck

Diffusers
LiteRT
ONNX
Safetensors
Model card Files
xet
 Community
cocccck / custom_nodes /ComfyUI-Roop /README.md
bbbboiwow's picture
bbbboiwow
Upload folder using huggingface_hub
edb09f2 verified
preview code
|
Raw
History Blame Contribute Delete
2.51 kB

ComfyUI Roop Custom Nodes

Custom nodes for ComfyUI to enable face swapping using the Roop library.

Features

  • Face swap between two images
  • Optional face enhancement
  • Batch image folder processing
  • Face swap for videos
  • Webhook sender for images (tensor) or file paths
  • Auto-increment filenames to prevent overwriting

Node Descriptions

RoopFaceSwap

  • Basic face swap using source and target image tensors
  • Returns a single swapped image tensor

RoopFaceSwapWithEnhancer

  • Like RoopFaceSwap but also runs a face enhancer after swapping
  • Useful for improving visual quality

RoopBatchFaceSwap

  • Takes a source image and a folder of input images
  • Applies face swap to each and saves results to an output folder
  • Supports enhancement and multi-face mode

RoopFaceSwapVideo

  • Swaps a face from an image into a video file
  • Keeps original FPS and frames
  • Saves a new video file

RoopSendWebhookImage

  • Sends an image tensor to a webhook URL
  • Accepts optional secret for HMAC verification
  • Ideal for live image results

RoopSendWebhookFile

  • Sends a file path (e.g. video or saved image) to a webhook
  • Supports HMAC-secured sending
  • Useful for post-processing delivery

Requirements

  • Python 3.8+
  • torch
  • torchvision
  • Pillow (PIL)
  • requests
  • roop (clone from GitHub)

Installation

  1. Clone this repo into your ComfyUI custom nodes folder:

    git clone https://github.com/glitchinthemetrix16/ComfyUI-Roop

  2. Install required Python packages:

    pip install torch torchvision pillow requests

  3. Clone the roop repo for Google Colab Environment:

    git clone https://github.com/FurkanGozukara/roop /content/roop

  4. Ensure that run.py exists in /content/roop

Installation For Google Colab With T4 GPU (Tested and working) 

%cd /content/

!git clone https://github.com/FurkanGozukara/roop
%cd roop

#Tested and updated 23 August 2023 commit
#!git checkout da1ef285f1d43bd0cc8b9cdb9a0f80f7ae793a97
!pip install onnxruntime-gpu && pip install -r requirements.txt
!pip install onnxruntime-gpu --upgrade
!apt-get update --yes
!apt install nvidia-cuda-toolkit --yes
!pip install opennsfw2 keras --upgrade
!pip install numpy==1.26.4

Usage Notes

  • All temporary files are saved inside roop_dir/temp_io
  • File overwrites are avoided via auto-renaming
  • Video swapping works best with shorter clips due to performance
  • Webhook nodes allow sending output to remote servers or APIs