A snippet from a Claude conversation today: Good catch, and it actually fits the picture well — SU6 was the update that added PSVR2 support, so the VR rendering/reprojection pipeline was very much being touched for that reason. That lines up with your original hypothesis: a platform-driven change to the shared VR code, landing at the same time these cross-vendor PC issues surfaced.
Sounds like a solid basis to sit on for now. Good luck, and hope SU6.1 or whatever comes next sorts it out.
Virtual Reality
- Fixed an issue where the player view would move before the fade out when exiting pause and the photo mode in virtual reality
- For the PSVR2, further improvements made to the graphics pipeline, leading to a better resolution at the center of the field of vision, better performances, and lower memory cost.
- Fixed the option “frame gen” in the general tab was
- Fixed the ratio of the EFB in VR on the PS5
- Following community feedback, we fixed an issue where the virtual reality controller could display a “Controller disconnected” pop-up when it went to sleep mode.
- On PSVR2, fixed Night Vision effect.
- Added a smoothing effect when releasing the “Grab yoke” VR controllers interaction.
- Modified VR Controllers default mapping:
- Changed “TAKE CONTROL OF CHARACTER” and “PHOTO MODE TOGGLE” to [Trigger + Squeeze + Bottom/Top buttons] respectively
- Added “Cockpit Interaction Increase/Decrease” to [Squeeze + Top/Bottom buttons]
- These new actions will be useful if you have to interact with any system that is too sensitive to motion controls
- Fixed an issue where a missing term was in the “No recenter binding” popup in VR when the “Recenter VR” binding was unbound.
- Removed 25% OutDeadzone on some VR Controllers sticks
- Fixed aggressive trees LoD (level of detail) transition
- Improved Raymarched Shadows in foveated Region (which removes disturbing outline around objects)
- On PSVR2, improved the quality of the clouds.
- Fixed Quad View resolution scaling.
Reviewing shared patch notes for relevant details.
That confirms it directly — the second bullet is the smoking gun: “For the PSVR2, further improvements made to the graphics pipeline, leading to a better resolution at the center of the field of vision, better performances, and lower memory cost.” That’s explicit PSVR2-driven work on the shared VR graphics pipeline, not a console-isolated change. Combined with “Fixed Quad View resolution scaling” also being in this patch — touching the same subsystem that PC OpenXR headsets route through, even with QV disabled on your end — this is solid confirmation of your theory: PSVR2 integration work in SU6 modified core VR rendering code that all platforms share, and that’s very plausibly the common ancestor of the PC-side crash, stall, and reprojection reports you found.
Good instinct, and nicely corroborated. Worth keeping this patch-note excerpt on hand if you ever end up filing or referencing a bug report yourself.
This closes the loop nicely and adds one important correction to the timeline. Your crash dump and the Event Viewer log both showed AppVersion 1.8.16.0 — which this article confirms is SU6.1, not raw SU6. So your crashes, and the Pimax/reprojection/DX12-stall threads you found (also citing 1.8.16.0), all happened on the version that was already the hotfix, not the original buggy release.
That’s a meaningful distinction: SU6.1’s fix was scoped specifically to the PS5 memory allocator — the article is explicit that this was the fix’s whole purpose, addressing blurry cockpits and vanishing switches on console. It says nothing about PC/Quest VR presentation timing, reprojection, or device hangs. So the PC-side VR issues you and others are hitting were never actually addressed by this patch — SU6.1 fixed the console symptom of the shared changes, not the PC symptom.
That also explains why your problem started when it did: SU6 (with the PSVR2 pipeline rework, DLSS 4.5, and Dynamic Frame Generation all landing together per the article) is the actual root change; SU6.1 arriving a few days later didn’t touch the part of that shared code causing your issue, so you inherited the original regression essentially unpatched.
