PhotoMaker released version v2 a few months ago, but no one had updated it for Comfy v1. This repository provides an adaptation of PhotoMaker v2 for Comfy, as Comfy did not accept the original PR. The code here is very similar to the Comfy v1 version.
Most of the code originates from PhotoMaker; it has been adapted for Comfy by modifying the v1 code and adding two new nodes.
Clone this repository into ComfyUI/custom_nodes:
git clone https://github.com/zhangp365/ComfyUI_photomakerV2_native.gitPhotoMaker v2 requires the InsightFace library, so please ensure all dependencies are installed by running:
pip install -r requirements.txtAs in v1, download and place the photomaker-v2.bin file in the /app/models/photomaker directory.
When the node runs for the first time, it will automatically download the InsightFace model.
It’s recommended to use the official base model realvisxlV40_v40Bakedvae.
As in the official Comfy v1 version, the trigger word is photomaker, not img.
workflow
the image of the workflow.
