search

Totoro97 / f2-nerf

Fast neural radiance field training with free camera trajectories
https://totoro97.github.io/projects/f2-nerf/
Apache License 2.0
933 stars 69 forks source link

blocky blurred area in demo #41

Closed lif314 closed 1 year ago

lif314 commented 1 year ago

In the demo video you gave, I found that there will be blocky blurred areas in some frames. This problem was even more pronounced when testing on my own dataset collected outdoors. What could be the reason for this? Thank you so much for your work!

image

On my datasets:

https://github.com/Totoro97/f2-nerf/assets/63114150/f753dd47-01b2-4516-ac80-2d46aee3fa23

Totoro97 commented 1 year ago

Hi, thanks for your interest in this project! The empty gray regions are octree nodes that are not visible to the training cameras, so they are automatically removed.

The video results look weird. I will first check if it is due to inaccurate camera poses. By the way, what is the split setting of training/testing sets?

lif314 commented 1 year ago

Hi, thanks for your interest in this project! The empty gray regions are octree nodes that are not visible to the training cameras, so they are automatically removed.

The video results look weird. I will first check if it is due to inaccurate camera poses. By the way, what is the split setting of training/testing sets?

I used the default settings, wanjinyou.yaml and wanjinyou_big.yaml.