search

Fictionarry / ER-NeRF

[ICCV'23] Efficient Region-Aware Neural Radiance Fields for High-Fidelity Talking Portrait Synthesis
https://fictionarry.github.io/ER-NeRF/
MIT License
985 stars 131 forks source link

About high-resolution data and other face tracker. #15

Open Xu-Zirui opened 1 year ago

Xu-Zirui commented 1 year ago

"Thanks for your code. Upon running it, I have 2 questions. First, when I use another face tracker, like MICA, I found the result is not good. Second, I tried high-resolution images (1024 * 1024) and the loss became NaN."

Fictionarry commented 1 year ago

Sorry we have not tested our code on other face models, and currently we only support resolutions about 512x512. You may have to make some modifications by yourself if you want to use the code in other cases.

RuojiWang commented 8 months ago

Sorry we have not tested our code on other face models, and currently we only support resolutions about 512x512. You may have to make some modifications by yourself if you want to use the code in other cases.

can i use your code for 128x128 resolution?