Image arrangement

[LHXhh]

Hi, I have a question for you. I used colmap to estimate the pose of the drone video frame, whose camera arrangement was parallel to the xy plane, and then I converted the code from the nerfstudio format to transfer it from colmap to nerfstudio, but for some reason, the overall image arrangement of the video frame was distorted, and it should be parallel to some plane. This causes the scene not to be found later when the mesh is extracted with box.


But I treated another drone dataset of the same scenario in exactly the same way, and its results were normal. There's something about this that I don't understand. I'd appreciate it if you could answer it.