Paul-Edouard Sarlin

It probably refers to [doc/features.md](https://github.com/cvg/pixel-perfect-sfm/blob/main/doc/features.md) but the docstring is incorrect, the output is of type `List[pixsfm.features.FeatureMap]`. cc @Phil26AT

Yes the refinement can improve the accuracy of the camera intrinsics.

Does [this PR](https://github.com/cvg/Hierarchical-Localization/pull/179) solve your issue?

This is now fully supported by hloc via https://github.com/cvg/Hierarchical-Localization/pull/210. An example is given here: https://github.com/cvg/Hierarchical-Localization/issues/219#issuecomment-1219213026

Thanks a lot Dmytro, this is very useful! Some of these issue are indeed odd but I don't have much experience with Clang either. I will look into this whenever...

I do not understand your question. Please provide proper logs as plain text and explain why you think that they do not conform to the expected behavior.

Closing due to inactivity.

If COLMAP is already installed, `pip install git+https://github.com/colmap/pycolmap` should work because it builds from source too.

Hi, thank you for your interest! I just found out about [this recent WACV2022 paper](https://openaccess.thecvf.com/content/WACV2022/papers/Leotta_On_the_Maximum_Radius_of_Polynomial_Lens_Distortion_WACV_2022_paper.pdf) that derives the maximum radius for more complex camera models. The solution is not trivial...

### 7Scenes It turns out that the default path pointed to the raw SuperPoint+SuperGlue SfM model, while the results reported in the paper are based on the dense depth maps...