model-viewer
model-viewer copied to clipboard
WebXR Depth Occlusion Support
Description
Seems this has shipped in Chrome a while ago and would bring model-viewer's WebXR closer to SceneViewer parity. https://chromestatus.com/feature/5742647199137792 I'm not sure if there's already a three implementation but thought it's worth opening this here. Might be related to some of the post-processing feature requests (bloom, AO, etc.).
OS
- [X] Android
- [ ] iOS
- [ ] Linux
- [ ] MacOS
- [ ] Windows
AR
- [X] WebXR
- [ ] SceneViewer
- [ ] QuickLook
I started working on figuring out how to put together depth occlusion + light estimation + hit test with WebXR. Is there a plan to support these things via Model-Viewer? How much is the project time frame?
depth occlusion hasn't been added for the same reason that xr-environment
is not on by default: it doesn't meet our quality bar. I actually find myself generally turning off the depth occlusion feature in Scene Viewer because it was confusingly wrong so often. Then again, I haven't tested in a while - has AR Core improved here?
I generally agree, but for context we do have some projects where clients inquire about depth occlusion (or rather, complain about the lack of it). It's not something that I'd want to have on by default for everything for sure, but project-specific it can make a lot of sense.
AR Core's depth occlusion unfortunately still isn't great. I think iOS went the better way there (they're only doing a depth-aware people mask, not general scene depth)...
PRs welcome!