Torsten Sattler

The SfM algorithm can work without SIFT, see for example https://github.com/tsattler/visuallocalizationbenchmark/tree/master/local_feature_evaluation Colmap requires a certain amount of feature matches between two images for them to be considered as an initial...

Do you want to create a pull request for this? Would be happy to merge this in.

If I remember correctly, we enforced the rig constraint when computing the reference model in Colmap. The rig bundle adjuster was experimental by that time, so some things might not...

@insikk Sorry for the late reply. This somehow slipped past my attention. What type of features are you using? I recommend using an upright Hessian affine detector (https://github.com/perdoch/hesaff) with RootSIFT...

I would assume that you get at the very least close to the results reported in the paper. In my experience, using the right type of features and a well-trained...

@insikk Could you give me some details on which OS you used? The provided files worked fine for me under Linux. It does not make sense to replace the existing...

In general, the intrinsics stored in the `.out` and `.nvm` are an approximation to the correct intrinsics: Both assume that the principal point is in the center of the image...

The map is used to compute the calibration. Without a map, this software is not able to calibrate images.

The database can be created using Colmap and the vocabulary can be created using the DBoW2 library.

Colmap has a good tutorial: https://colmap.github.io/tutorial.html Here is an example for using DBoW2: https://github.com/youkely/InfrasCal/blob/master/src/examples/train_voctree.cc