About preprocessing custom data

[kongsemancheng]

Hello. Thanks for your great work! I have a question about preprocessing custion data.
I have processed the data according to the preprocessing steps you provided, but I found that the scale_mat is quite different.
My code:
python imgs2poses.py ../../scan63 python gen_cameras.py ../../scan63
and my result:

I am so confused about this question. I might have encountered an issue with "scale_mat" during the preprocessing phase, and I'm seeking some assistance. Can anyone provide guidance or suggestions?

[sxbpwm]

Hello, I have encountered this problem as well. May I ask if there is a good solution?

[kongsemancheng]

Hello, I have encountered this problem as well. May I ask if there is a good solution?

Sorry for my delayed reply, I hope it's still helpful to you! I later realized that the camera_sphere.npz for NeuS on the DTU dataset was generated following the processing of IDR (https://github.com/lioryariv/idr/blob/main/DATA_CONVENTION.md) rather than obtained from Colmap, which results in differences in pose and scale.