Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

COLMAP GUI finds poses while nerfstudio COLMAP settings does not? #3137

Closed
abrahamezzeddine opened this issue May 10, 2024 · 2 comments
Closed

Comments

@abrahamezzeddine
Copy link

abrahamezzeddine commented May 10, 2024

Describe the bug
COLMAP via NerfStudio does not find enough poses although COLMAP GUI finds the poses correctly. Using sequential for this.

To Reproduce
Steps to reproduce the behavior:Perform ns preprocess data with COLMAP with sequential image frames. Fails.
Perform automatic reconstruction sparse data with COLMAP with high or low quality and sequential image frames. Successful.

Expected behavior
The nerfstudio preprocess data should pass accordingly, just like COLMAP GUI should do.

Screenshots
Skärmbild 2024-05-11 002303
Skärmbild 2024-05-11 002410

Additional context

@KevinXu02
Copy link
Contributor

Hi! The pose matching process in COLMAP is random, so you may need to try a few more times for it to match successfully.

@abrahamezzeddine
Copy link
Author

abrahamezzeddine commented May 13, 2024

Quite a time consuming process to be honest to do such guessing game... Especially for large datasets...

Currently using COLMAP GUI with high success rate actually. Then I import them to nerfstudio via this command below;

ns-process-data images --verbose --skip-colmap --colmap-model-path sparse/0 --num-downscales x --data folder/images --output-dir folder/. There is no documenation for this so I am just guessing this is how nerfstudio would import a complete model. Anything else that I might have overlooked with this command?

I need however to create two new folders, "sparse" and "0" and export the COLMAP model into folder 0 in order for nerfstudio to detect the model.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants