Denser sparse point-cloud

[mutunake]

Even though many matches are found, at the end of the Align Photos only few points are computed. For example for a dataset with 69 photos 2555240 matches found and 193386 tie points, only 64413 points are computed (~30%). Is there an option to reconstruct all of the found points? Can you explain please why only a part of them are used?

Thank you!

[James]

Tie points are the points in your photos that have been found to have valid matches with points in other photos. The larger number (2555240) is probably the number of points detected, or key points, which are either not matched, or not matched in a valid way.

Given that you have 193386 tie points, each of which is matched with at least one other tie point in another photo, the maximum number of points you can have in your point cloud is 193386/2 (96693), and the fact that you have fewer than this means that some of your tie points exist in more than 2 photos.

Basically it takes at least 2 tie points (which only exist in the 2D photos) to make a single point in the 3D sparse cloud.