search

generalizable-neural-performer / gnr

Implementation of "Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis"
178 stars 9 forks source link

selected views and distortion #6

Closed YuLiHN closed 2 years ago

YuLiHN commented 2 years ago

Hello!

Thanks for your incredible work! I have two questions:

  1. In the paper, you said only 4 views are given in the model for training, may I know which 4 views?
  2. In dataset Test10, I found camera distortion parameters in the npy files. Do we need to undistort the images before using, or the images are already undistorted so distortion parameters can be ignored?

Thanks!

SuX97 commented 2 years ago

Hi @YuLiHN :

Thanks for your question,

  1. Four views are mutually orthogonal, specified here
  2. Undistorting is needed for real-world captured data, and is performed here
YuLiHN commented 2 years ago

Hi @YuLiHN :

Thanks for your question,

  1. Four views are mutually orthogonal, specified here
  2. Undistorting is needed for real-world captured data, and is performed here

Thanks for your explanation!