cmusatyalab/mega-nerf

why the ray_altitude_range used by x

chufall opened this issue · 3 comments

chufall commented

I think the ray_altitude_range is used cutted to include the highest buiding 。
Did it align the Z axis ,and why it used to cut the X axis value in the code?

Thanks a lot
qc

hturki commented

Sorry, the coordinate system used in Mega-NeRF is pretty confusing, but the position coordinates are [down, right, bottom]. The idea is to further constrain ray sampling to not needlessly sample points below ground level.

chufall commented

Thank you for your reply!
I have been confused.
I have been visualized the cameras of the Rubble scene. The results show its cameras faced aligned X axis.
But After I processed my own dataset captured by my drone(DJI M300 with RTK) followed readme and issues (1. with pixsfm 2. colmap aligner 3. run the convert script ) . The visualized results show its cameras faced aligned Z axis.
Should I switch the X and Z value of the result before running the convert script?

  1. switch the X and Z aixs in the camera c2w matrix
  2. switch the X and Z value of the camera pos vector

Am I right? Is there any script ?
Thanks a lot!

lhc991025 commented

Thank you for your reply! I have been confused. I have been visualized the cameras of the Rubble scene. The results show its cameras faced aligned X axis. But After I processed my own dataset captured by my drone(DJI M300 with RTK) followed readme and issues (1. with pixsfm 2. colmap aligner 3. run the convert script ) . The visualized results show its cameras faced aligned Z axis. Should I switch the X and Z value of the result before running the convert script?

  1. switch the X and Z aixs in the camera c2w matrix
  2. switch the X and Z value of the camera pos vector

Am I right? Is there any script ? Thanks a lot!

Hello, have you successfully customized the dataset and trained it. Can you provide your method