Philipp Lindenberger

Hello, unfortunately I could not reproduce your problem on any of my machines, but it seems this issue is related to hloc/pycolmap, not pixsfm. Without further information, my best guess...

Hi, sorry for the late reply. Usually the weights should be downloaded automatically, but you can also manually get them from [https://www.dropbox.com/s/hnv51iwu4hn82rj/s2dnet_weights.pth](https://www.dropbox.com/s/hnv51iwu4hn82rj/s2dnet_weights.pth).

Hi, this is caused by an incompatibe pycolmap installation (usually the one from pypi). As written in the readme, pixsfm requires installing [COLMAP and pycolmap from source](https://github.com/colmap/pycolmap#building-from-source).

Hi @robotshanxi, thank you for reporting this issue! Without a full log it's hard to locate the problem, would you be able to share your log (with [debug](https://github.com/cvg/pixel-perfect-sfm/blob/main/pixsfm/__init__.py#L27) enabled), your...

Hi @robotshanxi, I guess the problem is that some features are missing for certain observations, but unfortunately I cannot reproduce the problem on my end. Would it be possible to...

Hi, the problem is that pybind fails to resolve the constructor here. You can either serve the points via the keyword argument: ``` image = pycolmap.Image("image.png", points2D=pts) ``` or use...

Hi @ThomasParistech, thanks for the PR! It looks great. Re pycolmap: Yes good idea, pycolmap's Track is still a bit clumsy to use.

Hi @ThomasParistech, indeed this use case is not supported atm, but would be great to have. I think this should be quite simple to implement, just make [`reference`](https://github.com/cvg/pixel-perfect-sfm/blob/main/pixsfm/features/bindings.cc#L258-L266) a dataclass...

Indeed meshing is currently missing. I added bindings to branch [meshing](https://github.com/colmap/pycolmap/tree/meshing). Note that Delaunay meshing is unlikely to be part of a future pycolmap release on PyPI since it requires...

Yes this is still in the pipeline to be merged, and should be ready now. Tested on Aachen v1.1 with setup [`loftr_aachen`](https://github.com/cvg/Hierarchical-Localization/blob/loftr/hloc/match_dense.py#L56-L69) (no superpoint keypoints), which should run on machines...