mw 
Only C$5.6 million of Lockheed Martin Canada’s C$11.7 million commitment to Avalon Holographics is new, it counts toward Canadian offset obligations, and the lower-cost OLED flat panel it funds still has no public date. One NOVAC table is in use at Lockheed’s Virginia innovation center.
Genesis Interactive says CoasterMania’s room-scan update runs on its MIT-licensed QuestRoomScan package. The game’s developer skipped Gaussian splats as too heavy for Quest, splat training still needs a PC, and an October 16 fix for a locale bug that discarded 964 camera records is on main but not in a release.
SlimeVR Server v22.0.0-rc.1 opens the public beta of a ground-up Kotlin rewrite with step mounting, toe tracking and bone routing. Autobone is not ported yet, two features wait on a new protocol, and pull request #1798 remains unmerged.
Halo firmware 0.8.20 traces a post-standby display and camera failure to a glitch on a shared I2C bus, adds 6 dB to phone audio, and trims echo. Brilliant Labs’ 1,000-cycle test covers one unit, and the notes give no count of affected glasses.
Godot 4.8 beta 1 freezes features with Apple Vision Pro hand tracking and PS VR2 Sense controller support on the engine’s standard XR nodes. It is still a beta with no stable date, and the visionOS module shipped for the Mobile renderer only.
Release 2610.15.2 of the R.E.A.L. VR framework adds GTA V Enhanced, four years after a Take-Two DMCA notice pulled Luke Ross’s original mod. It is early access, has no motion controls, needs BattlEye disabled, and needed a hotfix a day later.
Quest Browser 152.1 adds experimental WebGPU composition layers for WebXR behind a chrome://flags switch, after space-warp in 149.1 and foveation in 150.1. The W3C binding is still an Editor’s Draft, and developers are still falling back to WebGL.
A peer-reviewed Pitt case report describes the first skull base neurosurgery with Apple Vision Pro: no complications, but one video stream, 64 minutes of setup, a roughly 90-minute stock battery and no measured latency.
Museum of Ages labels each of its 17 stops documented, reconstructed, conjectural or not known. Only 14 of its 24 picture-room attributions are documented, and phones get filmed machines instead of live renders.
Tether’s 254M-parameter QVAC TranscribePsy runs speech recognition on phones and is pitched at smart glasses, but the release is batch-only, names no glasses partner, and its benchmarks are self-reported.