search

POSTECH-CVLab / SCNeRF

[ICCV21] Self-Calibrating Neural Radiance Fields
MIT License
463 stars 45 forks source link

Question about downsampling factor #7

Closed Len-Li closed 2 years ago

Len-Li commented 2 years ago

Hi, Thanks for sharing your work!

I am confused about the downsampling factor in the LLFF dataset. In your config file, downsample factor is set to 8. However, in the original nerf repo, downsample factor is set to 4 according to this. I am curious that why do not follow the original setting. This may be unfair for comparision.

jeongyw12382 commented 2 years ago

For the case of NeRF, our implementation is based on the PyTorch-NeRF, which is publicly available on https://github.com/yenchenlin/nerf-pytorch/blob/master/configs/fern.txt. If you visit the aforementioned link, the factor value is set to 8. We have followed the setup introduced in the repository to minimize code or setup modification that is already verified.

Thus, we re-evaluate the baseline model with our code for a fair comparison.

In addition, based on my experiments in the very early stage of the research, we have not observed large difference using factor 4 and factor 8. However, I have not verified whether there is a large gap between the two factor setups in the final model.

I hope this description might help you.

jeongyw12382 commented 2 years ago

Closing the issue. Please reopen the issue whenever you have trouble using the repository.