why settle for a few images from #
stablediffusion# when you can slowly walk your way around the sample space and create hyponotic videos you can't look away from? In this 2min video (~1hr to render on A100) I'm smoothly interpolating between random noise inputs into the model.