This collection of Workflows would be called iKki.
Currently the zip has 3 workflows,
1. Text to image with adetailers and upscaler.
2. Image to image with adetailers and upscaler.
3. ControlNet with upscaler.
Use this fork of comfyui-impack-subpack if you are having trouble with Adetailer models not working with pytorch >=2.6
https://github.com/RedsAnalysis/ComfyUI-Impact-Subpack.git
clone it in ComfyUi/custom_nodes
Adetailer model links are below
Mouth(segm) https://civarchive.com/models/1306938/adetailer-2d-mouth-detection-yolosegmentation
Hands(segm) https://civarchive.com/models/329458/hand-detailersegmentation-adetailer
Eyes(segm) https://civarchive.com/models/334668?modelVersionId=465360
Face(segm) https://civarchive.com/models/1076050/adetailer-anime-girl-face-segmentation
BBOX(Face & Hands) https://huggingface.co/Bingsu/adetailer/tree/main
Custom Nodes
rgthree-comfy
impact pack
impack subpack from the fork above
art-venture
Controlnet Aux
UlimateUpscaler
ComfyUI_essentials
etc.
Description
FAQ
Comments (9)
please make version for automatic 1111 .pt file pls,
I don't understand what to do with these text files, can someone help me?
they are Json files, extract the zip file that you download and just drag the json file on to a comfyui workspace, it will open up the workflow and then you can use it
Where can I find the breast detection models?
Best Adetailer Workflow of all time 🤍
i wanted to try this but im having trouble with the UltralyticsDetectorProvider it says it's not installed even though it is, i tried installing both with gitclone and the comfy mananger i'm running out of ideas, help please.
Had the same issue, apparently you have to install the Ultralytics plugin manually. pip install ultralytics in a command prompt is the fix I was told. Not sure if you have to do that inside comfy too due to vemu or whatever it's called
To clarify, yup, gotta go into vemu. So I went to \Documents\ComfyUI\.venv\Scripts in a command prompt, ran Activate, then did pip install ultralytics. Now it seems to be working.