Question: How important is it that source images are divisble by 8 pixels in both height and width?

[marcsyp]

I know some image sampling processes are most efficient in buckets of 8x8 pixels, so I'm wondering if providing an image to controlnet that is a few pixels off and has to be resampled to an 8-factor resolution has a significant impact on the result quality. Any thoughts?

Thanks.

[geroldmeisinger]

#224 (comment): "SD can be trained with any resolution as long as the number can be mod by 64." Also see #365