[nowittyusername]

I was using stable diffusion 1.5 when it came out and had my first LSD trip shortly after that, prolly like a few months. Anyways, what struck me was how similar the closed eyes visuals were on LSD compared to the generated images from stable diffusion when i was using it on low CFG and also some of my poorly trained textual inversions at the time. Watching the "training process" of the textual inversion in the early epochs made a lot of such images before the TI finally completed. Makes me think if the processes are somehow related, like if in human brains the reason we don't have experience seeing these "hallucinations" is because we have many robust subsystems that filter out the noise and make the mental model cohere on a stable world view.