Good news: this finally hits the quality I was looking for!
Bad news: ComfyUI broke their save model node so this model had to be broken into parts, each of the images has the layout for the model, simple instructions though for people who need to know what to do:
For the download, take the model and set it in the normal checkpoints folder.
For the clips, download both, name the big file big and the small file small and put it into the clip folder, not the clip vision folder.
For the VAE, put it into the VAE folder.
In ComfyUI, use your normal setups but with a small change:
load the model with a checkpoint loader and connect it to any model node you would normally.
Load the clip with a dual clip loader, set the big clip on top and small clip on bottom, link them into your prompt boxes.
load the VAE, set it to the VAE spot you would link with the model.
