search

yenchenlin / nerf-pytorch

A PyTorch implementation of NeRF (Neural Radiance Fields) that reproduces the results.
MIT License
5.37k stars 1.04k forks source link

Scene with known camera poses #45

Open mertkiray opened 2 years ago

mertkiray commented 2 years ago

Hello, thank you for this project :)

I got a problem while training on a scene.

When I extracted the camera poses from COLMAP this is the output

https://user-images.githubusercontent.com/19538303/142732551-c89f3ed4-06c9-4567-a4d4-d2d5e3272bd9.mp4

the result is not so crispy as the scene is a little bit hard to learn but

But when I use the camera poses from the dataset I got a result like this

https://user-images.githubusercontent.com/19538303/142732596-d1f68062-72f8-46f1-9608-36568be8fee7.mp4

The difference between the two is really great. I changed the gt camera poses to the coordinate system of right, upwards, backwards.

What am I doing wrong?

Thanks.

GUO-W commented 2 years ago

Hello, thank you for this project :)

I got a problem while training on a scene.

When I extracted the camera poses from COLMAP this is the output

kitti_exp_2_spiral_100000_rgb.mp4 the result is not so crispy as the scene is a little bit hard to learn but

But when I use the camera poses from the dataset I got a result like this

kitti_exp_pose_1_spiral_025000_rgb.mp4 The difference between the two is really great. I changed the gt camera poses to the coordinate system of right, upwards, backwards.

What am I doing wrong?

Thanks.

Hi, have you found out the reason, please? I got similar bugs when training on my scene. Thanks!

mertkiray commented 2 years ago

Hey @GUO-W, the problem was with the camera coordinate system I was using, after applying the correct transformation, it is fixed now.

Ben-chen-alt commented 2 years ago

Hey @GUO-W, the problem was with the camera coordinate system I was using, after applying the correct transformation, it is fixed now.

how to trans the world coordinate using the stand nerf

DRosemei commented 2 years ago

the world coordinate using the stand nerf

@mertkiray How do you translate kitti gt pose to NeRF-style "pose_bounds.npy"? Thanks in advance!

LEEJUHUI6635 commented 1 year ago

Hello, @mertkiray. I had a problem same with you. How could you refine your depth images or depth scales? How could you get a correct max depth and min depth?