WebXR

[m-schuetz]

TODO

• [ ] Speed not optimal for small-scale scenes.
• [ ] EDL rendering shows nothing on the Quest.
• [ ] Annotations behaving weird. When clicking on an annotation in the sidebar, the annotation title in the scene vanishes. Happens in annotations.html, but not in features_ca13.html Update: Looks like only annotations with html elements behave weird?

done

• [x] Skybox y and z are flipped
• [x] Better VR-UI. E.g., show scale while scaling, add exit-vr button, allow movement with either controller, ...
• [x] Adaptive sizes broken in VR
• [x] Only does default rendering atm. Missing EDL and HQ rendering. (resolution: stick with default rendering for now)
• [x] Check potential new errors, e.g., using wrong WebGL context (1 vs 2), update to three.js 123 works with all functionality, etc.
• [x] Only works if background = skybox
• [x] Navigation, including scaling
• [x] Gracefully enter VR and exit VR, make sure the viewpoint after a switch correlates to the viewpoint before the switch.
• [x] ONLY WORKS WITH CUSTOM CHANGES TO three.js!!! three.min.js doesn't work. Many demos include three.min.js! Make it work!
• [x] chiller.html not working because THREE.LineGeometry missing?
• [x] Measurement tools not working
• [x] (near-clip) Frustum culling is broken because the scene is scaled via camera.scale.

[midnight-dev]

I've admittedly dragged my feet on implementing a suitable pseudo VR for systems not able to use the WebXR spec... Sorry about that 😞

I'll be able to jump back in after the holidays. If I recall, Rooi had a fantastic looking AR demo. I'd like to help wherever I can. It would be great to have a WebXR branch to target.

By the way, making the rendering engine compatible with the current version of Three shouldn't be too hard - just kind of tedious. Should that be the first goal for this milestone? All Potree users would benefit from that even if they never use VR. It's been a hot minute since r109 was released, and the amount of bug fixes / performance improvements in each revision of Three is impressive.

[midnight-dev]

Sweet Christmas, I completely missed that you already made a WebXR branch and checked off a bunch of goals...

Welp, never mind my previous comment. At this rate, you'll be finished by the time I get to the office.

By the way, I think the last objective about EDL is a duplicate since EDL and HQ both require the fragment depth extension. Which reminds me... Are you using something akin to Google Cardboard (any phone or tablet really) to test your new feature? That would explain why the alternate renderers don't work.