Scaling factor inconsistencies

[Chirobocea]

I have set up this code as they indicated in description. I was trying to upscale images from AFLW with a factor of 5. However, even that I had scaling flag set to 5 in command line there were few images in the final output that was scaled with factors like 5,81; 5,13; 8,11 (usually slightly bigger, I also do not remember a new actual scaling factor that was less then the selected one).

I used inference_codeformer for this.

The images that I used are typically 450x450, sometimes padded with 0 and saved in .jpg format.

Do you have any idea why this happened?