search

jixinya / EVP

Code for paper 'Audio-Driven Emotional Video Portraits'.
297 stars 49 forks source link

Confusion about audio2lm/test.py #23

Closed zizizihua closed 1 year ago

zizizihua commented 1 year ago

audio2lm/test.py#L113 example_landmark = example_landmark - mean.expand_as(example_landmark)

Both example_landmark and mean are loaded from config['mean'], which will output 0?

jixinya commented 1 year ago

Hi, we are trying to predict the landmark motion displacement for the example_landmark in test.py, so that we can set the example_landmark with different identities or mouth shapes. If the example_landmark is mean, which is the case in test.py, the input for the encoder will be 0.