LODGroup behaves very differently in Daydream on device than in the editor. It seems to select *far* lower LOD levels at the same distance on device than in editor. I've seen some whispers of this for Gear VR, where people are simply doing things like setting their QualitySettings.lodBias = 3.8 when they're not in the editor. What's the correct answer here, though? Is this a bug? Or is this expected since the display is in stereo and the screen size of objects is actually smaller on each display, even though it would be larger if the entire display were a single eye? Thanks!
I don't know anything specific here in this regards, but you can't compare in Editor to in Device, at least not for Daydream. The in editor experience it totally mocked by Google unless you are specifically using one of the non-HMD devices in the Standalone Player settings.
But the work flow for using lodgroup is to drag the dividers between the levels of detail, which causes the scene view camera to visibility zoom out until you've reached the desired size/distance for that lod level. This is the scene view, not the game view, so no stereo, etc at this stage. This then translate really well to similar size/distance at runtime across various screen resolutions and mobile/desktop/web platforms, but falls completely apart in vr. Why is that? And whatever the reason, how are we to properly use this component for platforms or daydream, where what we see during the work flow described above at authoring time bears no resemblance whatsoever to the results at runtime? How does one know what parameters to select at authoring time when the visual controls for this thing simply lie in the editor? By trial and error? Or is there some lod bias number that should simply correct it for daydream? Thanks!
In all honest, I know nothing about LOD group behavior and what it is/is not supposed to do. Are you using 5.6 or 5.4 GVR13? If 5.6, you might want to cross post this into the Beta forum list as someone from the larger Unity community might be able to provide more help than just I can. If it repros in 5.6 please file a bug report and post the case ID. That way it should route to the correct team.
I will create a simple repro project and file a bug report, then post it here and in 5.6. Thanks for offering to help, joejo!
It's submitted as bug case number 879100 and it has a very simple repro project attached, including an already-built APK of the project to try on Daydream. The bug report includes all of the details about what to observe in editor vs daydream headset.
Thanks. I'll take a look as I can today and see if I can determine anything but I am hoping someone from the wider audience can provide a better idea or understanding.