Question about filtering low-quality generated samples

[capitaso]

Thank you for your excellent work.

I noticed that the README mentions filtering out low-quality generated samples. However, I could not find details about the filtering procedure.

Could you clarify:

1. Was the filtering manual or based on quantitative criteria?
2. If the latter, any objective metrics (e.g., reconstruction error, similarity measures, or diffusion-related scores) were used?

Understanding this would be very helpful for reproducing and extending your method.