Here is the workflow I use almost everytime for my image using Chain Sampling with ClownShark Sampler and only upscale with ClownShark Sampler too.
Pretty easy to use and only need 4 custom nodes (For V1)

For V1

https://github.com/pythongosssss/ComfyUI-Custom-Scripts

https://github.com/rgthree/rgthree-comfy

https://github.com/ClownsharkBatwing/RES4LYF

https://github.com/edelvarden/comfyui_image_metadata_extension

For V2 (you need the previous one)

https://github.com/ltdrdata/ComfyUI-Impact-Pack

https://github.com/yolain/ComfyUI-Easy-Use

https://github.com/jags111/efficiency-nodes-comfyui

https://github.com/chrisgoringe/cg-use-everywhere

https://github.com/ltdrdata/ComfyUI-Impact-Subpack

https://github.com/giriss/comfy-image-saver

https://github.com/alexopus/ComfyUI-Image-Saver

https://github.com/rcsaquino/comfyui-custom-nodes

For V3 (you need the previous one)

https://github.com/shiimizu/ComfyUI-TiledDiffusion

Model for control net on Upscaler

https://huggingface.co/lllyasviel/sd_control_collection/blob/main/diffusers_xl_canny_full.safetensors

Note for the Face detailer

I added a different model and VAE loader for the face detailer, you can either use the same model and baked VAE, or change it for whatever you prefer.

VAE I used for the Face detail (for some reason I get better result with it)

/model/1684460/illustrious-xl-v20

Note for the Upscaler

Here you can change the Upscale ratio, you can change it to 1.5 or whatever you want, you need to write a*1.5 on both box to work.

Here you can change the Tiled VAE decode if the upscale take too long or you get OOM error, the lower the number faster it get but will change the image output.

You can also change the Sampler Name for a linear one (like Euler) or a multistep one (like Dpmpp_2m). If you change it by another one you mostly need to add more steps and try differents scheduler.

Exponential one are the the slower (especially res_5s) but have better result IMO.

Leave step_to_run at -1, it means the sampler will do all the steps.

PLS tell me if I made a mistake on the Upscaler nodes with all the Color Match and JPEG Removal it can get pretty confusing to link them.