Error running `python imgs2poses.py data/testscene`

[darthgera123]

Im getting this error, when trying to do manual installation

Need to run COLMAP
Features extracted
Features matched
ERROR: Failed to parse options: unrecognised option '--output_path'.
Traceback (most recent call last):
  File "imgs2poses.py", line 18, in <module>
    gen_poses(args.scenedir, args.match_type)
  File "/home/gera/Desktop/CVIT/Research/code/LLFF/llff/poses/pose_utils.py", line 268, in gen_poses
    run_colmap(basedir, match_type)
  File "/home/gera/Desktop/CVIT/Research/code/LLFF/llff/poses/colmap_wrapper.py", line 71, in run_colmap
    map_output = ( subprocess.check_output(mapper_args, universal_newlines=True) )
  File "/home/gera/anaconda3/envs/llff/lib/python3.7/subprocess.py", line 411, in check_output
    **kwargs).stdout
  File "/home/gera/anaconda3/envs/llff/lib/python3.7/subprocess.py", line 512, in run
    output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['colmap', 'mapper', '--database_path', 'data/testscene/database.db', '--image_path', 'data/testscene/images', '--output_path', 'data/testscene/sparse', '--Mapper.num_threads', '16', '--Mapper.init_min_tri_angle', '4', '--Mapper.multiple_models', '0', '--Mapper.extract_colors', '0']' returned non-zero exit status 1.

[dilaratank]

This means you don't have the images in a separate folder called 'images'. This is necessary. I have stumbled upon the same problem a lot!

[nitthilan]

I too am facing the same problem. Can someone explain what these data structures mean? pts3d, poses in the save_poses function

[SunM00n]

I have another traceback on the same script imgs2poses.py

Traceback (most recent call last):
File "imgs2poses.py", line 18, in
gen_poses(args.scenedir, args.match_type)
File "/content/LLFF/llff/poses/pose_utils.py", line 276, in gen_poses
save_poses(basedir, poses, pts3d, perm)
File "/content/LLFF/llff/poses/pose_utils.py", line 66, in save_poses
cams[ind-1] = 1
IndexError: list assignment index out of range

I'm not sure why the index goes out of range..

[robclouth]

@fishirenee did you solve this? one of the images hasn't registered correctly, could it be that?

[SunM00n]

@robclouth I revisit the dataset. I just deleted the poses that are barely blank with minor features (such as an image of the white ceiling). I noticed those poses caused this type of bug.

@fishirenee did you solve this? one of the images hasn't registered correctly, could it be that?

[robclouth]

Is there anyway to figure out which images are causing the problems? Colmap is dropping one, (the number of poses is 15 but there are 16 images), but I don't know how to figure out which one it is.

[robclouth]

It would be nice if the code just ignored images with no pose instead of failing

[kajalsanklecha]

anyone with the following issue?

`Need to run COLMAP

Features extracted

Features matched

Sparse map created

Finished running COLMAP, see data/frame16/colmap_output.txt for logs

Post-colmap

Cameras 5

Images # 15

ERROR: the correct camera poses for current points cannot be accessed

Done with imgs2poses`

How do I solve this? The poses are not generated.

[ofekp]

I figured it out, see the line

LLFF/llff/poses/colmap_wrapper.py

Line 64 in c6e27b1

'--output_path', os.path.join(basedir, 'sparse'), # --export_path changed to --output_path in colmap 3.6

I simply changed the code from output_path to export_path and it worked
I will add that I installed colmap on Ubuntu 18.04 at the time of writing this comment and the installed version was colmap 3.4 which requires the change I mentioned above.

is it possible to support both versions in the wrapper code?

[shreeramsigdel77]

@kajalsanklecha Did you solved the issue that you have mentioned above?

[longbowzhang]

@kajalsanklecha Hi, I am facing exactly the same problem as you, do you have any idea to solve it?
Thanks in advance.

[starhiking]

I have fix the issue at #60

The issue is because some views(images) are not registered in the colmap, so we cannot use pose_boundary.npy to process all views(images)

[Yes-Jumby]

thanks，good done！

[JunJie1213855]

@robclouth I revisit the dataset. I just deleted the poses that are barely blank with minor features (such as an image of the white ceiling). I noticed those poses caused this type of bug.

@fishirenee did you solve this? one of the images hasn't registered correctly, could it be that?

could you please provide some code to solve it?