Feature suggeston: Stable Diffusion Only Option

[jasonmhead]

What could be cool is to add the option to only use Stable Diffusion to generate an image as well, like you can with Karlo, so it can be a dual interface

[kpthedev]

I'm currently looking into a more efficient way to use the upscaler. Right now, it uses a lot of GPU memory which is a major limiting factor. I'm exploring tiling the input image and upscaling in smaller chunks. If I can get that working reliably, this feature will be easily implemented.