HoloLens 2 VIEW reference space and spatial graph node space origins

[emaschino]

Hello,

When creating a VIEW reference space or a spatial graph node space to track HoloLens 2 PV camera [1], where is physically located the space "origin" on HoloLens 2 device? Is it still the left front VLC camera of the rigNode [2]?

Then, when locating HoloLens 2 VIEW "camera" space (using any of the above) in application space (Microsoft unbounded reference space) with xrLocateSpace, does the returned pose has to take into account the extrinsic transform from camera to rigNode origin, as was the case with the deprecated Perception API (code excerpt in Using Media Foundation section) [3]?

From my experiments, origins of VIEW reference space and spatial graph node space are different. But the observed difference doesn't match the camera's extrinsic transform, so can't really say where's the origin located.

Thanks for any input/suggestion/clarification.

[1] https://github.com/microsoft/OpenXR-MixedReality/issues/88#issuecomment-883545132 [2] https://arxiv.org/abs/2008.11239 [3] https://learn.microsoft.com/en-us/windows/mixed-reality/develop/native/locatable-camera-cpp