I may have modified a few stable diffusion files to speed up the image creation process. Seems to produce more accurate results which might not be desired in some of the models I use. It does not prohibit me from recreating the exact image that I've previously created with the model, so I may just keep two versions of it - and do a clean install for my other models.

Modified the upscaler_utils.py file. Which I didn't think would have an effect on regular images. But I fear it has.