search

RenYurui / PIRender

The source code of the ICCV2021 paper "PIRenderer: Controllable Portrait Image Generation via Semantic Neural Rendering"
Other
509 stars 63 forks source link

About inference speed of Audio-Driven Facial Reenactment #4

Open forest520 opened 2 years ago

forest520 commented 2 years ago

Could you tell me what is the inference speed of Audio-Driven Facial Reenactment on Nvidia GPU？ Is it possible to execute in real-time such as 30fps? Thanks!

RenYurui commented 2 years ago

Hi, Thanks for the interest. We have not performed such estimations before. However, you can first try to estimate the speed for the image generation part using the provided code. It is better to modify the code to make different frames operate in parallel to increase the speed. Yurui

liushengtong-nreal commented 2 years ago

Hello, I have the same problem now, do you know the inference speed of Audio-Driven Facial Reenactment on Nvidia GPU now?Could you please share your conclusions, I would be grateful if you could.