Changed workflow and reuploaded!

My workflow for ComfyUI with process similar to Forge webui (txt2img -> 2-pass highres fix (1st pass: foolhardy remacri upscale, 2nd pass: denoising after upscale) -> facedetailer (face and eyes).

Adapted for NoobAI-Xl V-pred-based checkpoints (rescale CFG enabled with 0.3, you can change).

If you use epsilon-based models (pony, Illustrious, NoobAI-Xl Epsilon 1.1) you can bypass rescale CFG node or remove it because epsilon-pred models don't need this node.

If you want to add more lora nodes make more nodes and add one after another.

It is my test workflow which I made by myself to make similar process which I do it forge.

Unfortunately I didn't make img2img / inpaint process and Civitai metadata nodes (too difficult for me). Also maybe some settings can be more optimal in facedetailer but it is too complicated for me - I used adetailer in forge all time so don't judge too much.

Also my Facedetailer only workflow - https://tensor.art/workflows/923327030492139408 (If you did highres fix but need to fix face. For example if you did pic on civitai and don't want to use their terrible facefix).