Replies: 2 comments
-
This can be realized through the img2img: by feeding a smaller image - which is similar to what the Highres. fix does. That does make a two step process through: generate the smaller image and then enlarge with second sampling method. I don't know of a way to save the latent space. So, I don't think that one option is covered in the two step method. |
Beta Was this translation helpful? Give feedback.
0 replies
-
Reduce the denoising strength, I think that's the issue you are having |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I don't know if this is possible but I'll just explain what I would love to see.
If I enter steps 50 on the "sampling steps" slider and enable the highres. fix, I want to have the option to either just use the firstly generated image and then stop the generation or be able to set how many sampling steps the AI should do after the highres. fix was "done".
As I said, I don't know if that is possible, but if so, that would be awesome.
My model generates very good images out of the highres fix and it only gets worse when it samples afterwards.
If there is already an option for that, I'd be very happy if you could tell me. ^^
Beta Was this translation helpful? Give feedback.
All reactions