AliaksandrSiarohin

In log/..../00000##.pth.tar

I guess you just have to run the demo with your ckp and see if you get satisfactory quality. Usually I stop training when first value perceptual no longer decrease.

You out of memory. Did you resize all your videos to 256x256? How many gpus you are using?

Log is in the same folder as the vox checkpoint.

You can save all the models + kp_driving in memory. For each source you could also save encoding of the source.

1. I mean load the models only once for all videos, if you already do it than nvm. 2. It saves the frame into a list in order to save...

.cpu() transfers data. You can try to create a pytorch tensor t and save all the frames there, then use io.mimsave('q.mp4', t.cpu().numpy()) maybe this will be faster.

There are something problems with last frame, just skip it.

You should run the command that is outputted.

What is the script output?