[Question] running with custom images on pre-trained model

[chris-aeviator]

Thanks for releasing ADOP. If I want to test ADOP on my own image dataset, I'll need to

a) pre-process them with Colmap (do I need a dense reconstruction, a sparse reconstruction or just the camera poses?) b) convert them to ADOP format via colmap2adop c) load them into the ADOP/scenes folder

before running ADOP?

[darglein]

Hello,

a) You need both, the dense and sparse reconstruction. However from the dense reconstruction only the point cloud is required.

b) yes, also consider running preprocess_point_cloud after colmap2adop

c) This will automatically be done by colmap2adop.

After that run adop_viewer on you scene to check if the point cloud / poses are correct.

[chris-aeviator]

Thanks for the quick clarification.