>>58418330I'm currently using ComfyUI to generate images with fuwafuwamix1.5 and counterfeitXL (both from CivitAI). The workflow produces garbage if I change any of the settings slightly so I assume I've somehow hit some sweet spot in using SDXL to prompt from the latent image produced by SD1.5, although I could be completely incorrect here. I don't know what I'm doing, to be honest with you, but it works so I'm not complaining.
https://i.postimg.cc/NMggYYm3/Comfy-UI-00281.pnghttps://i.postimg.cc/RFn2jDbk/Comfy-UI-00331.png