-
Is that a bug? what should I do? |
Beta Was this translation helpful? Give feedback.
Answered by
mashb1t
Mar 21, 2024
Replies: 1 comment 2 replies
-
This is not unusual as Stable Diffusion is only trained on specific 1024x1024 and similar sum of pixels aspect ratios. |
Beta Was this translation helpful? Give feedback.
2 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
A potential solution would be to use upscaling as one of the last steps in your generation process or decrease the weight / choose another inpainting mode for inpaints on upscaled images.