Generating point cloud data about zind

[ivaylob-zg]

Thank you for your interest in ZInD!

We don't have official experiments on that, but this is certainly something you can try on your own, e.g. using off-the-shelf SfM pipelines like OpenMVG or OpenSfM that support sparse spherical cameras as an input. This will probably give you some sparse localization and point clouds for a subset of the scenes with enough capture density and a rich set of features to match. However, we expect this to fail quite often in the majority of the scenes that have wide baselines (between the cameras) within empty, featureless rooms.

[ywyue]

Thanks for your reply! It is clear now.