vicgalle/stable-diffusion-aesthetic-gradients

Cant reproduce with diffusers

hadaev8 opened this issue · 1 comments

https://colab.research.google.com/drive/1BShA8RtOCIcXtokprNoQa-QqmbA_Q-c3?usp=sharing

Here is colab, with shifted text encoder it produce very different output.
Any suggestions?

Figured it out.