mabel-lr
commented
1 year ago using the code in this folder: https://github.com/Fyusion/LLFF/tree/master/llff/poses, create the following script ->
from pose_utils import gen_poses
gen_poses(colmapdir)where colmapdir is the directory where the output of the previously run colmap is, specifically the folder sparse (the output was colmap_sparse, change the name to sparse)
i did a slight modification to pose_utils -> change the defintion of the gen_poses function by:
def gen_poses(basedir, match_type=None, factors=None):
extract poses_bounds.npy, I have seen something about the LLFF format, there is a possibility of doing it with COLMAP or without COLMAP, doesn't matter how we do it but we have to get this .npy to train on our images