I just completed training with 128px and now I was thinking on converting with pics at 512px.
How would I do that? merging the aligment files of both 128 extracted faces and 512 with the alignment file from the 'A' video?
I think you're misunderstanding the process. The extract resolution has no bearing on conversion.
The converted faces will be converted at the size that they are in the frame.
My word is final
I have an aligment file with frames at 128px for training.
I merge that file with the aligment file from the video to have a single alignment file.
In conversion that file uses frames at 128px despite the fact that the original frames could be at 512 pix or more.