Performance in VR (Lagging issues in Oculus Quest 2)

[bittu97]

Lagging issues are there even though I followed the VR Performance instructions provided on the readme of the GitHub link.

1. #define MAX_NUM_STEPS to 100 (also tried by tuning MAX_NUM_STEPS of other functions too)
2. and deleted this line: #pragma multi_compile DEPTHWRITE_ON DEPTHWRITE_OFF

It starts lagging when it is rendering the Model, I observed, if the model is not inside the camera field range, the application is working fine, and when the model is in front of us it starts lagging. If we go closer to the 3D-Model it lags increasingly and the application goes out of control.

[mlavik1]

Hmm, I don't have an Oculus Quest 2 (or any VR headsets at the moment) to test with, so it's difficult for me to be helpful here, since I can't do any profiling..
It's expected that performance is worse when looking closely at the model (since more pixels are visible then), but the last time I tried with a VR headset the performance was still quite Ok. That was not a standalone headset, like Quest 2, and I had a relatively good GPU.

One thing I expect would help is to reduce the screen resolution, but it will look pixelated then.
Do you also have poor performance with the other rendering modes? (isosurface, maximum intensity projection)

[bittu97]

Hmm, I don't have an Oculus Quest 2 (or any VR headsets at the moment) to test with, so it's difficult for me to be helpful here, since I can't do any profiling.. It's expected that performance is worse when looking closely at the model (since more pixels are visible then), but the last time I tried with a VR headset the performance was still quite Ok. That was not a standalone headset, like Quest 2, and I had a relatively good GPU.

One thing I expect would help is to reduce the screen resolution, but it will look pixelated then. Do you also have poor performance with the other rendering modes? (isosurface, maximum intensity projection)

Yes, I have tried the maximum possible ways from my side. Also tried changing the transfer functions by tuning density and opacity, but nothing works.

[mlavik1]

Then I'm not sure.. Sorry!
I'm hoping to get a VR headset I can test with soon, so I can do some performance profiling for VR.

Only think I can think of is: If it helps to turn down the screen resolution, you might be able to get a performance boost by using DLSS. But AFAIK Unity only supports that in HDRP render pipeline, and I'm not sure if the shader is going to work with that out of the box. (see #130 )

[mlavik1]

Hi again!
So it seems like the project works well with HDRP (just add HDRP package, and do auto-upgrade). Not sure about VR though.. but if it works there as well, you could give DLSS a try!
https://docs.unity3d.com/Packages/com.unity.render-pipelines.high-definition@12.0/manual/deep-learning-super-sampling-in-hdrp.html

[bittu97]

Hi again! So it seems like the project works well with HDRP (just add HDRP package, and do auto-upgrade). Not sure about VR though.. but if it works there as well, you could give DLSS a try! https://docs.unity3d.com/Packages/com.unity.render-pipelines.high-definition@12.0/manual/deep-learning-super-sampling-in-hdrp.html

Hi buddy,
For now, I implemented a workaround by adjusting the near-clipping plane of the camera.

I'll try that too. I have one more doubt, as it is working quite decent on VR Devices, after that I tried building the app on SmartPhones too and the application keeps stopping.

[mlavik1]

Ok, thanks for the feedback!
I'll try building it on Android to see.

[bittu97]

Ok, thanks for the feedback! I'll try building it on Android to see.

ok thanks, I am confused that the 3D Model displayed is a runtime display of data points or it's a static mesh generated after scanning the raw data.

[mlavik1]

Ah yes, so it's not a static mesh :) It's real-time raymarched volumetric data.