Open toji opened 1 year ago
Hololens has two different modes of mesh creation. You can choose to do an initial scan and cache the data. If you during the experience, there are issues with the completeness of the mesh, developer can restart scanning. The second option is to continously scan, which is very expensive.
It makes sense to add method the session (ie startMeshDiscovery
/stopMeshDiscovery
) and an attribute that indicates that this is possible.
It would also make sense to add this to the planes API.
/tpac Should we allow for an api for mesh updates? Should we combine it with a planes API?
I know this doesn't apply to the current Meta implementation, but considering a more dynamic implementation like those seen in Windows Mixed Reality or Magic Leap, it's my impression that the mesh generation is somewhat expensive. I'm curious if there's generally expected to be a way to start and stop it on command, so that you could, for instance, do a "setup" phase for an application where you allow the device to establish the environment geometry, then click "Done" and turn of scanning to "freeze" the current meshes and allow the application to devote more of the device power to the app.
I think some investigation would need to be done to determine how feasible/useful that is.