Automatic boundary for guardian from depth map too small
Hello,
great that the depth sensor is now used to propose a boundary.
However, while this is OK for some use-cases, it really needs to be improved latest when working together with scene data, space sense and the like.
I tried to place touchable objects on planes of walls and the like, but the guardian kicks in before I can really touch that.
The reason is that the boundary is a very round shape with notable distance from anything around and on the floor.
Also it does not persist when you turn around while mapping and loses details it already had.
- Please, first, ensure that it does not forget details it already learned and perhaps also let in learn on the go to improve further while walking around.
- If there are objects detected surrounded by the floor, allow to ignore them and just have space sense for them/ extra planes in the scene.
When you have some moveable chairs or the like it is really hard to place something with some size dynamically in the boundary. - When one adds furniture near to the boundary, please ask whether one wants to align both, so the boundary ends exactly before the furniture/ plane.
- Last, please allow multi-room. Really bad if people cannot stay in VR/ MR to get something to drink in the kitchen.