Just follow the instruction in the workflow
Description
I've tested on Realistic and semi-realistic/anime and it works on them. ComfyUI will download a model for CLIPSeg masking node that detects the face when you first queue a prompt.
FAQ
Comments (2)
When loading the graph, the following node types were not found:
ImageToContrastMask
Sorry for not responding quicker, the node can be found here:
https://civitai.com/models/24869/comfyui-custom-nodes-by-xss