no usrさんのイラストまとめ

apparently i just had a bug in my training code & the 768 model WAS actually learning the style :facepalm:

time to tweak hyper-parameters...

This is fine-tuned from the #StableDiffusion2 base model (512px outputs).

For this model, I focused on portraits, but it should be able to generalize to full-body.

I will continue to work on a 768px fine-tune & also full-body & non-anthropomorphic fine-tunes.

@socalpathy @arkhan_voyager

@karan4d

#robodiffusion
#stablediffusion
#deforum

some friends from the latest checkpoint

#StableDiffusion

I wrote this plug-in with accessibility in mind.

That means, as long as your computer can run #krita, you can use this :)

(e.g. leveraging google colab as your backend server)

It’s yet another way to leverage the img2img capabilities of stable diffusion