TQTQliu / MVSGaussian

[ECCV 2024] MVSGaussian: Fast Generalizable Gaussian Splatting Reconstruction from Multi-View Stereo
https://mvsgaussian.github.io/
MIT License
397 stars 19 forks source link

Discussion on colmap #57

Open ZhenyuSun-Walker opened 4 weeks ago

ZhenyuSun-Walker commented 4 weeks ago

Dear sir, now when I try to run the python lib/colmap/imgs2poses.py -s <source path> with the correct and high-quality images (number is 20), the result is shown as below.

Sparse map created
Finished running COLMAP, see dataset/20240928-224635_scene_kitchen_20/colmap_output.txt for logs
Post-colmap
Cameras 5
Images # 3
Index out of range: ind=8, len(cams)=3
ERROR: the correct camera poses for current points cannot be accessed
Done with imgs2poses

Why it produces such result? Actually I can succeed in completing this process when I use other dataset, including your examples as well as my own customed datasets.

5p6 commented 3 weeks ago

I have the same problem,the dataset that i use just could not pass the colmap progran.

Need to run COLMAP
Features extracted
Features matched
Sparse map created
Finished running COLMAP, see dataset\colmap_output.txt for logs
Post-colmap
Cameras 5
Images # 2
ERROR: the correct camera poses for current points cannot be accessed
Done with imgs2poses
TQTQliu commented 2 weeks ago

The error caused by running the colmap script is due to low overlap, which can also be found https://github.com/Fyusion/LLFF/issues/36#issuecomment-648018595.

You can first run colmap sparse reconstruction https://colmap.github.io/cli.html# , and then run the script lib/colmap/imgs2poses.py to get poses_bounds.npy.