Torsten Sattler
By default, the `point_triangulator` will ignore the image_ids that are in the "empty" reconstruction and will use the file names of the images to get their corresponding image_ids from the...
Empty as in does not contain any 3D points (see my first reply). I typically just use numbers 1 to N for the image_ids.
The most common explanation is that either the poses or the camera intrinsics are not accurate enough. How did you obtain the poses?
I have no experience with Arkit poses, but Arcore poses are typically not that accurate. I assume the same holds for Arkit poses. You can try to relax Colmap's parameters...
You can try this pull request #2145
Maybe you can consider adding xfeat? I have been playing around with aliked and xfeat. In general, aliked seems to be better. But if there is no GPU, xfeat is...
We experimented with different learned features in the context of SfM. Will send you the current state to your demuc email. I am personally using xfeat with lightglue on my...
I have seen issues with similar structures before with XFeat (e.g., the Barn scene from the Tanks & Temples benchmark). Using Lighterglue / Lightglue for matching resolved this issue for...
> 1. Is there a way in COLMAP to add gravity information as a constraint in sparse reconstruction? No, there currently is not. > 2. How can I use the...
Please have a look here: https://colmap.github.io/faq.html#geo-registration You will need to define camera positions. Relative position constraints currently cannot be used.