Torsten Sattler

It seems to me that these are just warnings (that certain functions and definitions are deprecated) and not an error.

Looking at your example, the two 3D points do not seem to share any 2D point. The SIFT keypoint detector can detect multiple features at the same 2D position as...

Is this the image size after dense reconstruction, i.e., the size of the undistorted images?

As part of the dense reconstruction process, Colmap removes radial (and potentially tangential) distortion from the images. The size of the undistorted images (and the depth maps since they are...

The output of colmap are the images, cameras, and points3D files. You can extract poses and intrinsics from the images and cameras files, respectively. Once you have the reconstruction, you...

If you have the absolute poses, you can get the relative poses between any pair of images as T2 * inv(T1), where Ti is the transformation from world to camera...

Yes, the qvec and the tvec should give you the rotation and translation direction of the relative pose. For the definition of both and the way they are stored in...

Where exactly are you trying to read the values from? Can you provide details about what you are doing?

You can double-click on the cameras to see a pop-up window with the poses. But why not export the model and read the poses from the file? To get the...

These naturally should be the same as the poses in the .txt file if it is the same reconstruction. As mentioned above, you can compute the world coordinates of the...