search

InterDigitalInc / latent-transformer

Official implementation for paper: A Latent Transformer for Disentangled Face Editing in Images and Videos.
https://arxiv.org/abs/2106.11895
Other
143 stars 22 forks source link

inputs is not defined #4

Closed ucalyptus2 closed 2 years ago

ucalyptus2 commented 2 years ago

for the inference.py modification suggested, inputs variable is not defined, rather input_cuda var is defined but that is not valid for entry into net(...)

Please look into it and maybe some change in the instruction could be read.

cc: @Xu-Yao

Xu-Yao commented 2 years ago

Hi, I guess you are mentioning the encoding part using the PSP encoder. The modification that we proposed is just to save the latent codes generated by psp, to facilitate the video manipulation process. The two variables are already defined in the original inference.py file.