Clarifying Vimeo dataset for training

[teaandscones]

Hi,

I'm trying to replicate the training procedure and I am currently facing a couple of issues:

1. For Stage 1 training (10 epochs, cascaded, 256×256), which dataset did you use? Was it the septuplet subset or the original Vimeo videos?

2. Regarding the Vimeo dataset itself, which videos from the original Vimeo list were actually used, and how were they processed/prepared for training?

Thank you for your help.