search

zhanglonghao1992 / One-Shot_Free-View_Neural_Talking_Head_Synthesis

Pytorch implementation of paper "One-Shot Free-View Neural Talking-Head Synthesis for Video Conferencing"
Other
764 stars 143 forks source link

Early training behavior #23

Closed GLivshits closed 2 years ago

GLivshits commented 2 years ago

Hi. First of all, thanks for your code. I'm training the model from scratch. I've found out that images from the generator look similar to the source image, but not the driving (pose, emotion, and so on). Perhaps, I'm at the start of training (only a single epoch has passed). Is this behavior expected? Please find the log image attached. Images_epoch_0_iteration_2000_rec jpg_00000000

zhanglonghao1992 commented 2 years ago

@GLivshits Yes, just wait a little longer. By the way, please run git pull to get the new code.

GLivshits commented 2 years ago

Ok, thanks:) But how many iterations should pass (approximately)? I am using 128 resolution with a batch size of 24.

zhanglonghao1992 commented 2 years ago

Ok, thanks:) But how many iterations should pass (approximately)? I am using 128 resolution with a batch size of 24.

Emmmm, maybe 40-50 epochs?

GLivshits commented 2 years ago

Thanks!