Open
Description
Currently the code assumes the frame of reference is stage. For OpenVR that basically equates to not doing any stage transforms for methods like XRPResentationFrame.getDevicePose.
If the site requests eye-level or head-model, however, treating the frame of reference as stage (the default) can cause weird parallax issues in the resultant math (i.e. the pose position being doubled).
Implementing XRFrameOfReferenceType awareness should fix these issues for sites using non-stage frames of reference.