VR Discussions, Questions & Feedback

Monaco tarmac using H264/940Mbs via link cable looks great. Monaco Tarmac using AV1/200Mbs via Virtual Desktop looks crap. So, no AV1 is not necessarily better. Higher bitrate H264 looks better.

 

Thank you. So you (sometimes?) use Virtual Desktop? I only use Link cable because I thought it’s always better quality and I don’t have to care about reloading batteries. Is there any reason/advantage for using Virtual Desktop in stationary games?

 

I have a headstrap battery that I can plug into the USB port when using Virtual Desktop that charges my headset at a much better rate than my link cable. Virtual desktop also has some extra options to play with like SSW which is performed by headset chip, higher resolutions, sharpening, color management. But... You open up a different can of worms when going wireless (router to PC setup etc) so if you can get your link cable settings tuned up (higher bitrate) and the cable charges the headset enough to reduce the power drain, then thats good enough I would say.

 

You see I was assuming they would have integrated AV1 in the tool by now so apologize for the confusion
Its acessible on VD so I was thinking it was there in ODT too
Its difficult to follow META approach to features for end-users, sometimes

 

Just had an incredible 24 hours race at Le Mans, really fantastic experience. I started last in LMDh class and finally (thanks to some traffic) ended up first. Day night cycle is really cool, unfortunately that the framerate dips so much causing stutter. Any way you guys deal with this? Any settings to minimize the stutter at least?

Unfortinately enabling motion reprojection does not work at all on my graphics card using openxr, it decreases fps from 90 to 20 or so, Not sure why, never been perfect but this is really atrocious. Used to be better.

Playing on 6900XT, 5800X 32GB.

I use openXR FSR, 75%. whith MSAA medium. This nets me a 90fps at daytime, but dips to 65-70 in night time.

 

I recently changed my approach in AMS2 due to the same dips at night and the horrible stuttering that came with it

I do this on both my Q2 and NEO3 link

I push the rendered resolution and decrease MSAA to low then ignore upscaling in openxrtoolkit but focus on FRR instead when i drive with the pico headset

Or adjust FOV in the ODT tool when i use the META device

Its stable at 90fps doing so most of the time except in some non optimized tracks

 

As I understand it, the GPU motion vector and encoding hardware Motion Reprojection relies on is much weaker/slower in AMD cards compared to Nvidia, so it's not uncommon for Motion Reprojection to actually make things worse from a frametime/fps point of view. Same goes for other headset equivalents (ASW, SteamVR Motion Smoothing etc) to the extent they rely on the same hardware

Edit: You could try and switch between the two Motion Reprojection Types in OpenXR Tools for WMR (original/enhanced) to see if it makes things better.

 

OpenXR Tools for WMR is for.....WMR, not Meta/Oculus, so you don't want/need that installed.
Your (mis)use of the terminology suggests you haven't quite worked out how all the moving parts fit together.
Rather than rehash it all here though, for oculus/meta to work with OpenComposite you need to convince it to run in SteamVR mode, and for that you'll need this GitHub - rodolZ80/bypass-oculus-detector: Windows Library that simulates not having the Oculus service running and having the VR Headset disconnected.

 

Heard this before, but the performance hit is too big for that.

I read somewhere that openxr for windows is interfering with openxr toolkit. It was suggested to uninstall openxr for windows. I'm gonna try that.

Also it used to work much better before, so it is clearly a bug.

 

Funny you should mention that, I have a G2 and suffered similar minor tracking issues this weekend, never seen them before in AMS2.
Thing is I also did a couple of flights in MSFS and the tracking was perfectly fine, same light conditions. I wonder if the sensors are picking up the monitor mirror perhaps (which I have windowed), might need to investigate further.

 

Chuck a towel over the monitor and test

 

I agree that it does seem to be too dramatic a difference, even for that.
However, I'm not sure what you mean by OpenXR for Windows interfering with OpenXR Toolkit. the OpenXR for WMR app is just a UI that lets you change a few WMR options. You still need the WMR runtime. OpenXR Toolkit is just an OpenXR layer, it can't do anything without an OpenXR runtime (in this case WMR). If anything, it is the thing that is 'interfering'. There may be some conflict, and you could disable the toolkit temporarily to see if you get the same behaviour. You could also enable the WMR frame timing overlay to see if it tells you where the delay is - GPU or CPU.
Good luck either way - I hope you're able to get to the bottom of it. If you do, please update us here to help out anyone else who hits the same issue.
Cheers

 

I completely agree with the conflicting info and I also don't understand how it would/could work . Just saw the post on Reddit with some people saying it solved their issue. I will try later this week and give feedback here.

 

Would shutting off the monitor work for not overheating it, or it's also a matter of avoiding reflections. I could do both...

I have a Lenovo Explorer VR, and its inside-out tracking is maybe too primitive and somedays it drives me nuts.

 

Shadow causes stutters

I have Quest 3 and using Oculus runtime. I play at 72hz with 1.5x supersampling in Oculus App. I noticed some microstutters recently and recorded a 2 laps race at Imola with 13 cars in the grid.

PC Specs: AMD Ryzen 7600, RTX 4070Super, 64GB RAM at 6000mhz

Test run with shadows: Everything is maxed out except grass and particles. The weather is sunny. As its seen, neither CPU or GPU bottlenecks the system. Yet, the frametimes are not stable at all. There are microstutters and time to time, there are stutters as well.


Test run without shadows: Every setting is the same except the shadows are OFF. The weather is sunny, too. The big stutters are gone and microstutters are decreased. The load on the GPU and CPU didn't change that much.


Comparison: The lows are much better without shadows. Thus, less stutter happened.


Is this behaviour normal even thou there is no GPU and CPU bottleneck in both cases? What is the matter with the shadows that can cause this much stutter? Could it be LoD issue?

 

Nice to see some real numbers and analysis here. Given there are no obvious usage bottlenecks from gpu or cpu (max thread is the one to look at here, rather than total) then it's probably waiting for something, but that can be hard to track down without detailed application traces. Could be io, could be memory (RAM or VRAM). Not sure if CapFrameX can help here.
Also worth noting, that tools like CapFrameX have to use a certain amount of guesswork as they don't have full visibility of what the application/game is doing. In particular, there is always a certain amount work that goes into generating each frame that isn't captured by the hooks it uses.
Hopefully Reiza can look at this, reproduce and maybe discover an as yet unknown problem or optimisation to improve things.

 

I checked the RAM and VRAM usage and those were fine as well. Even though there was a wait, it wasn't because of insufficient resources. VRAM usage was around 7.5 GB which is way lower than what 4070super has. (12GB)

To be honest, there are still microstutters even without the shadows. It's better than before but it's still there. I hope someone from Reiza have a look to it.

 

I had the odd microstutters, changing render frames ahead from 1 to 3 in performance settings fixed it for me. I run a 12 core amd processor with a 3090 to a G2 headset.

 

Thanks, I'll try it out. Have you noticed any input latency with this change?

 

Render frames ahead 3 helped a bit but it didn't eliminate all the microstutters. I also disabled the Windows Defender but no luck. I think something is wrong with the game engine in VR, especially shadows, which are problematic. But even without shadows, there should be some other setting that causes microstutters.