VR tips and tricks - Improving your VR experience! -Updated 9th/Dec/2020

[BH_Adabadoo_VR]

Hmm, been busy lately haven't had time to read here much.

 

Caught up looks like VRSS is a no go but still excited to try FSR soon.

 

Got my new computer with 3090 but still packed in a box.  Not sure when I'll have time to set it up.

 

Cheers all!

[Jan097]

Since IL-2 switched to deferred shading with 4.006, VRSS shouldn't work anymore.

[SCG_Fenris_Wolf]

On 8/17/2021 at 9:29 AM, KG_S_Jan097 said:

Since IL-2 switched to deferred shading with 4.006, VRSS shouldn't work anymore.

Exactly, VRSS and eye tracking won't properly work, unless IL-2 goes DX12 or to forward rendering. They'll have to go with time anyway and update it continuously, the game and its expansion stack is very much like a service.

 

The fact that they have not fixed the game's VR integration for the Vive Pro 2 in 3 months and for Q2 via Virtual Desktop in 9 months and real world size based on dynamic IPD handed over by OpenVR lets me doubt any progress in the regard to the VR implementation though. 

 

 

I know about FSR, I'll add it soon. I must run comparisons with it regarding spotting. I'm also looking for a way to disable hard v-sync in the Vive Pro 2, to switch it to adaptive.

[SCG_Redcloud111]

10 hours ago, SCG_Fenris_Wolf said:

I'm also looking for a way to disable hard v-sync in the Vive Pro 2, to switch it to adaptive.

Hey, Fenris, I haven’t heard you mention this v-sync issue or even how it affects the VP2. Can you elaborate what the issue is and why adaptive is better?

Edited August 19, 2021 by SCG_redcloud111

[BH_Adabadoo_VR]

15 hours ago, SCG_Fenris_Wolf said:

Exactly, VRSS and eye tracking won't properly work, unless IL-2 goes DX12 or to forward rendering. They'll have to go with time anyway and update it continuously, the game and its expansion stack is very much like a service.

 

The fact that they have not fixed the game's VR integration for the Vive Pro 2 in 3 months and for Q2 via Virtual Desktop in 9 months and real world size based on dynamic IPD handed over by OpenVR lets me doubt any progress in the regard to the VR implementation though. 

 

 

I know about FSR, I'll add it soon. I must run comparisons with it regarding spotting. I'm also looking for a way to disable hard v-sync in the Vive Pro 2, to switch it to adaptive.

So I tried it quick today.  Here are my results:

 

G2 went from SS50% to SS60% to offset .77 expect gain with some increased final res to target.

Used .5 radius and .77 default.

 

Went from 46 fps to 52 fps on my little test mission I always use with lots of clouds.

 

So with 8% increase in scaled final target gained 13% fps rendering much less pixels and it looked pretty good to me for a first test.

I'll probably go to SS70% or SS80% to get my render back closer to where it was and scale to even higher final target to get to original fps and see how that goes in future.

 

Looks good in center but off center was more blurry than normal.  Will try radius at 2.0 next time and see what performance is like.

I'll be using it on my old machine for sure.

 

Let us know what you think for spotting.  Very curious.

 

Cheers!

Edited August 19, 2021 by BH_Adabadoo_VR

[SCG_Fenris_Wolf]

7 hours ago, SCG_redcloud111 said:

Hey, Fenris, I haven’t heard you mention this v-sync issue or even how it affects the VP2. Can you elaborate what the issue is and why adaptive is better?

VP2 at 90Hz target settings always keeps 90Hz on its displays. If the PC cannot serve 90fps, it will be set to 45fps + Motion Reprojection. Motion Compensation is optional (known as Motion Smoothing to WMR, or as Async Space Warp to Oculus).

 

Adaptive V-Sync would be that it also serves 85fps, or 80fps, or 70fps, and then ramps back up to 90fps. This induces motion sickness in most people, not in me though. Hence I'd prefer that.

 

 

2 hours ago, BH_Adabadoo_VR said:

So I tried it quick today.  Here are my results:

 

G2 went from SS50% to SS60% to offset .77 expect gain with some increased final res to target.

Used .5 radius and .77 default.

[...]

 

Cheers!

Hey mate,

 

if you mention performance tests, please use absolute pixel numbers, e.g. 2600 x 2600. SSxx% differs from headset to headset, sometimes from machine to machine. So we cannot transmit information to others like this, unfortunately.

 

But I read you had a performance improvement? That's quite nice! 

 

Cheers! 

[SCG_Redcloud111]

8 minutes ago, SCG_Fenris_Wolf said:

VP2 at 90Hz target settings always keeps 90Hz on its displays. If the PC cannot serve 90fps, it will be set to 45fps + Motion Reprojection. Motion Compensation is optional (known as Motion Smoothing to WMR, or as Async Space Warp to Oculus).

 

Adaptive V-Sync would be that it also serves 85fps, or 80fps, or 70fps, and then ramps back up to 90fps. This induces motion sickness in most people, not in me though. Hence I'd prefer that.

Yes, I would prefer this. I didn’t realize this was related to V-sync. I miss being able to lower SS settings on Pimax and locking in 90fps. I would settle for even 65fps if it stayed there and didn’t start reprojection. I prefer the occasional stutter to the blurring and ghosting.

 

Thanks for the clarification.

Edited August 19, 2021 by SCG_redcloud111

[SCG_Fenris_Wolf]

With a target framerate at 90Hz and the PC just pushing 65fps of information, you also get the same blurring and ghosting in other headsets + plus stuttery head movement. Just saying. That situation has always been suboptimal. 

 

For me it has always been either 90Hz or 45Hz + Motion Compensation, in the Reverb G2 as well. I think the problem arises from the clouds + shadows at ultra high resolutions. We didn't use 3100 x 3100 or 3500 x 3500 before the G2/VP2. 3080ti/3090 cannot push such resolutions in IL-2 at 90Hz. Either we lower it and use FSR as well, or we ramp it up and play at 45Hz (or 60Hz in VP2) with motion compensation...or abandon shadows/cloud quality radically. Same applies to DCS and other ported VR games by the way. I don't think there is an alternative to this at the moment.

Edited August 19, 2021 by SCG_Fenris_Wolf

[SCG_Redcloud111]

35 minutes ago, SCG_Fenris_Wolf said:

With a target framerate at 90Hz and the PC just pushing 65fps of information, you also get the same blurring and ghosting in other headsets + plus stuttery head movement. Just saying. That situation has always been suboptimal

I mean, I would be happy targeting 65fps if I could get the VP2 to avoid any reprojection or blurring. I just tried DCS with Steam's fixed fps at 45 and it was much less reprojection when turning and looking at the ground. But I still saw ghosting when dogfighting. (In DCS I left Steam on 150% SS and used VP2 Ultra. I lowered DCS rendering to 8. Visuals looked good enough ... but again, the headset still ghosted.) I have yet to dogfight in the VP2 without ghosting.

Edited August 19, 2021 by SCG_redcloud111

[dburne]

I have recently switched running my Vive Pro 2 from Ultra (90Hz) to Extreme (120Hz).

So far I am pretty happy with it. I run it at around 3100x3100. Have been flying a DCS Hornet campaign with it and really liking the results.

My fps will fluctuate from 40 fps on the carrier to 60 fps whilst airborne. But it is smooth and not much ghosting at all. At least not enough to be annoying, if there is I hardly see it.  I do not use Vive's motion compensation either, I do not like it at all.

 

I am about done with that campaign and will try it in IL-2 also in my PWCG campaign with the 120 Hz refresh rate. I will probably crank it up to around 3200 to 3400 per eye for IL-2 and see how it does.

[SCG_Redcloud111]

Dburne. I haven't tried Extreme in DCS. I use it in MFS 2020 where ghosting doesn't matter and I am not moving my head around quickly. But I am trying to go in the other direction from high fps that drops into reprojection all the time. What bothers me is that when I move my head quickly, or look at the ground while in a turn, or try to follow an air contact, the reprojection destroys the clarity. When my head is still and I am just flying, it is very smooth. Also, I try for 3100 in most games, except MSFS

[SCG_motoadve]

3 hours ago, SCG_Fenris_Wolf said:

Adaptive V-Sync would be that it also serves 85fps, or 80fps, or 70fps, and then ramps back up to 90fps.

How do we turn ON Adaptive-V-Sync?

[dburne]

6 minutes ago, SCG_redcloud111 said:

Dburne. I haven't tried Extreme in DCS. I use it in MFS 2020 where ghosting doesn't matter and I am not moving my head around quickly. But I am trying to go in the other direction from high fps that drops into reprojection all the time. What bothers me is that when I move my head quickly, or look at the ground while in a turn, or try to follow an air contact, the reprojection destroys the clarity. When my head is still and I am just flying, it is very smooth. Also, I try for 3100 in most games, except MSFS

 

Strange - do you have motion compensation turned off in the Vive software?

[SCG_Redcloud111]

Dburne, no, I only use compensation in MSFS, even thought it causes artifacts. I think I am just very sensitized to blurred movement because I had my Pimax 5k+ dialed in very well at around 2000 vertical pixels at 90fps in IL2. There was no ghosting or blurred movement. I am a big fan of the VP2. I just wish we could completely turn off reprojection.  

Edited August 19, 2021 by SCG_redcloud111

[dburne]

21 minutes ago, SCG_redcloud111 said:

Dburne, no, I only use compensation in MSFS, even thought it causes artifacts. I think I am just very sensitized to blurred movement because I had my Pimax 5k+ dialed in very well at around 2000 vertical pixels at 90fps in IL2. There was no ghosting or blurred movement. I am a big fan of the VP2. I just wish we could completely turn off reprojection.  

 

When I was running 90 Hz, I would see framerate of 30, 45, 90. At 30 ( usually on deck of super carrier) I would see stuttering/ghosting. At 45 or 90 the experience was very good. Now at 120 Hz, I see 40 and 60 fps. And with both experience is very good. Which is why so far I am liking the 120 Hz setting better.

I will be getting back to my IL-2 PWCG campaign either tomorrow or Saturday and will be interesting to see how it does in it. I expect it to even be a little better in it so may crank the res up a little higher.

[SCG_Redcloud111]

So, I have to admit at Ultra (120ghz), the reprojection was happening but it was much smoother. I did see slight ghosting with other aircraft, but I wasn't noticing the ground blurring. I spent several hours playing around in the K4 and the Spit. I think I may actually try to do the Epsom campaign now that it seems to be smooth enough to be enjoyable and actually look good.  

 

Now, if only Necksaver could start working in DCS ...

Edited August 20, 2021 by SCG_redcloud111

[dburne]

1 hour ago, SCG_redcloud111 said:

So, I have to admit at Ultra (120ghz), the reprojection was happening but it was much smoother. I did see slight ghosting with other aircraft, but I wasn't noticing the ground blurring. I spent several hours playing around in the K4 and the Spit. I think I may actually try to do the Epsom campaign now that it seems to be smooth enough to be enjoyable and actually look good.  

 

Now, if only Necksaver could start working in DCS ...

 

Yeah I have been very happy so far with my Vive Pro 2 running at 120 Hz.

As long as I can hold 40 or 60 fps the experience is very good. If it drops to 30 fps then stuttering starts.

I have it dialed in now for DCS to be able to maintain 40 fps. Previously at 90 Hz I could not maintain 45 fps at least on the carrier, it would drop to 30 fps and stutter. Hope to be done with the campaign I have been working on sometime today, and tomorrow get back into testing IL-2 with it at 120 Hz in PWCG.

[SCG_Redcloud111]

I am waiting for them to fix the distortion on IL2. It is way too noticeable when I turn my tilt my head left and right.

[BH_Adabadoo_VR]

Saw this for quest2 distortion issue.

Was hoping someone with new VP2 could try various IPD at startup and see if distortion is gone if set in middle??

 

quote below:

 

Hi,

 

I have only ever played Il2 via wifi  on Virtual Desktop and this has always been a problem for me but I have today, by accident, found a fix on my quest 2.

 

I have found that if I start VR with the pupiliary distance at the max position (68mm) which is correct for me the effect is very noticeable, however, if the headset starts up in the medium setting the effect is not visible and does not come back if I push it out to max PD.   If I take the headset off with max PD set and it goes to sleep then put it on again, the probelm is back and is not fixed by putting the PD to medium. I need to close VR, put it on medium PD and start VR again.

 

I don't know why this should work, but it does.   Today was a good day to fly. 

Operating System
    Windows 10 Pro 64-bit
CPU
    Intel Core i5 3350P @ 3.10GHz
    Ivy Bridge 22nm Technology
RAM
    16.0GB Dual-Channel DDR3 @ 798MHz (11-11-11-28)
Motherboard
    Gigabyte Technology Co. Ltd. B75-D3V (Intel Core i5-3350P CPU @ 3.10GHz)
Graphics
       2047MB NVIDIA GeForce GTX 1060 6GB (ZOTAC International)  

 

Hope this helps.

[SCG_Redcloud111]

Ada

 

Either I didn't follow your instructions correctly, or this doesn't apply to the VP2. Didn't work for me. Truth be told, I wasn't fully clear on exactly what you did. If could write them out step by step, it might be helpful. I would try again.

[BH_Adabadoo_VR]

11 hours ago, SCG_redcloud111 said:

Ada

 

Either I didn't follow your instructions correctly, or this doesn't apply to the VP2. Didn't work for me. Truth be told, I wasn't fully clear on exactly what you did. If could write them out step by step, it might be helpful. I would try again.

So it wasn't me

 

Someone with the quest2 that has the same distort issue the VP2 has claimed that if they start game with the IPD centered the distortion isn't there.

If they start in any other setting it's there.

If they change IPD after game start it stays as started there or not there.  It seems to be related to IPD setting.  And at start not something that is fixed on fly.

 

So try to set VP2 as close to middle IPD as possible when launching game and see if it helps you too?

 

I have not tried myself in quest 2.

 

Thanks...

 

 

[SCG_Redcloud111]

Yeah, tried it. No worries. 

[SCG_Fenris_Wolf]

Same here, tried it, doesn't work either

[BH_Adabadoo_VR]

21 hours ago, SCG_redcloud111 said:

Yeah, tried it. No worries. 

thanks for trying was hopeful

[SCG_Fenris_Wolf]

- 06.09.2021 Update

- Included fholger's FidelityFX SuperResolution (FSR) Upscaler and how to use it.

- Reworked SteamVR Settings for each headset in light of FSR

- Adjusted Vive Pro 2, Reverb G2/1, Samsung Odyssey, added Index and Vive Pro --- ingame settings

- Re-added Pimax 8KX

[BH_Adabadoo_VR]

Hi Fenris,

 

Thanks for the hard work and the update.

 

I've been using FSR and trying different settings with it.

I came across results I thought I'd share that you could check and and see if you have similar results.

 

I'm now running on an i9-11900kf oc'd to 5.4 with 32GB RAM and an ASUS 3090.

 

I have my settings cranked up pretty high.

 

As a baseline running my G2 without FSR at 100% 3164x3096 I get 55fps on a mission with lots of clouds, motion smoothing off.

With the same settings on my G2 at 68% 2668x2552 I get 69fps on same mission with lots of clouds, motion smoothing off.

So I targeted 2552 as my resolution for FSR and scaled down to that render res for multiple tests of much higher % in SteamVR.

 

I record video of each test mission and use a few things I look at to test quality of the rendering in stills from the video recorded.

 

In each case I was truly rendering 2668x2552 regardless of the output scaled image as checked in the logs.

So I'd test 150% 200% 300% all the way to 500%.

 

G2 resolutions for me and fps results

7072x6920 500% fsr scale .368497  = 2668x2552 @59fps

6328x6192 400% fsr scale .411822  = 2668x2552 @60fps

5480x5360 300% fsr scale .475746 = 2668x2552 @63fps

4472x4376 200% fsr scale .582724 = 2668x2552 @64fps

3872x3792 150% fsr scale .672468 = 2668x2552 @67fps

3164x3096 100% fsr scale .823643 = 2668x2552  @68fps

 

Each tested producing several stills looking at the same things.  The stills are overlaid labeled in photoshop so I can turn them on and off look at the differences.

The higher the target resolution the better the result was in photoshop even though the scale # was getting very small for FSR.  The input render was always the same so the fps was close but more scaling work had to be done.  Still for a bit of extra scaling with the same input render the output is better the more it scales in the stills.

 

It's very hard to see with the naked eye but going from 100% to 200% by this method I feel is definitely noticeable to the eye and is more noticeable in stills.  I don't think the results above 200% are worth the drop in FPS but I'm sticking with 200% and a lower scale.  The 200% with scale .582 is significantly better than 100% with .823 both rendering 2668x2552.  In all cases I used .75 sharpness and .75 radius. I really liked testing originally with the debug on to get a sense of the radius area and also to see the latency cause by the scaler in the log files.  Doesn't .4 radius cause softness in the edge of view on the Vive Pro 2?  I thought the edge to edge sharpness was one of things you loved about it.  I'd have thought .4 radius caused too much softness off center.  I know when I used .5 default I thought the edge in G2 was too soft but 2.0 was much too much latency.  I settled on .75 for now.

 

I'm curious if you also try and see if you like 200% better using the same render res you have now.  Check the log to see what you are rendering and try 200% with a smaller scale to render to the exact same size in the log and see if you don't think the result is better.

 

In all the tests the only thing changing is the final target the input render res is the same.  It scaling to a bigger target that then oversamples into the headset produces a better result in the tests I've done without too much overhead. 200% seems to be the sweet spot.

 

End result for me is fantastic visuals with the settings cranked so high but only 60ish fps on very cloudy maps.

80-90fps on cloudless maps.  So motion smoothing on for cloudy maps sometimes and maybe off when no clouds.

But even at 80-90fps there's a juddery motion with motion smoothing turned off.  So it's like a choice between silky smooth and artifacts once in a while or judder even at 80+fps...

 

Thanks...

 

 

 

 

 

Edited September 6, 2021 by BH_Adabadoo_VR

[SCG_Redcloud111]

3 hours ago, BH_Adabadoo_VR said:

 

I'm now running on an i9-11900kf oc'd to 5.4 with 32GB RAM and an ASUS 3090.

Damn!

[halz]

I am somewhat surprised you're seeing the best results with FSR without using any in-game anti-aliasing.

 

AMD's FidelityFX/FSR docs rather explicitly say to only use FSR when a high-quality anti-aliasing is used earlier in the graphics pipeline.  https://github.com/GPUOpen-Effects/FidelityFX-FSR/raw/master/docs/FidelityFX-FSR-Overview-Integration.pdf (slide 14). They go so far to suggest games present the feature selection UI to only allow FSR to be used if/when a high-grade AA (MSAA 8x or TAA) option is selected.

 

After writing this, I realize IL2's AA is likely happening effectively in post-processing due to its usage of deferred rendering. Nvidia has some non-trivial guidance on doing MSAA effective within a deferred render pipeline https://docs.nvidia.com/gameworks/content/gameworkslibrary/graphicssamples/d3d_samples/antialiaseddeferredrendering.htm  So I guess what you're experiencing with FSR pushing a huge resolution into the headset is effective turning into super-sampled anti-aliasing, and still somehow the benefits of this for the sake of improved spotting are outweighing the drawbacks of FSR on a non-AA'd scene.

[TCW_Brzi_Joe]

5 hours ago, BH_Adabadoo_VR said:

In each case I was truly rendering 2668x2552 regardless of the output scaled image as checked in the logs.

So I'd test 150% 200% 300% all the way to 500%.

 

G2 resolutions for me and fps results

7072x6920 500% fsr scale .368497  = 2668x2552 @59fps

6328x6192 400% fsr scale .411822  = 2668x2552 @60fps

5480x5360 300% fsr scale .475746 = 2668x2552 @63fps

4472x4376 200% fsr scale .582724 = 2668x2552 @64fps

3872x3792 150% fsr scale .672468 = 2668x2552 @67fps

3164x3096 100% fsr scale .823643 = 2668x2552  @68fps

Same result you can have easier; to set your G2 resolution always 68% (2668*2552), and to change FSR scale to:

2.71

2.43

2.10

1.716 (that is your 200%)

1.50

1.21 (100%)

[BH_Adabadoo_VR]

9 hours ago, halz said:

I am somewhat surprised you're seeing the best results with FSR without using any in-game anti-aliasing.

 

AMD's FidelityFX/FSR docs rather explicitly say to only use FSR when a high-quality anti-aliasing is used earlier in the graphics pipeline.  https://github.com/GPUOpen-Effects/FidelityFX-FSR/raw/master/docs/FidelityFX-FSR-Overview-Integration.pdf (slide 14). They go so far to suggest games present the feature selection UI to only allow FSR to be used if/when a high-grade AA (MSAA 8x or TAA) option is selected.

 

After writing this, I realize IL2's AA is likely happening effectively in post-processing due to its usage of deferred rendering. Nvidia has some non-trivial guidance on doing MSAA effective within a deferred render pipeline https://docs.nvidia.com/gameworks/content/gameworkslibrary/graphicssamples/d3d_samples/antialiaseddeferredrendering.htm  So I guess what you're experiencing with FSR pushing a huge resolution into the headset is effective turning into super-sampled anti-aliasing, and still somehow the benefits of this for the sake of improved spotting are outweighing the drawbacks of FSR on a non-AA'd scene.

I see my best results with fxaa4x maxed out.  MSAA is still too expensive...

7 hours ago, TCW_Brzi_Joe said:

Same result you can have easier; to set your G2 resolution always 68% (2668*2552), and to change FSR scale to:

2.71

2.43

2.10

1.716 (that is your 200%)

1.50

1.21 (100%)

Interesting I'll try that, I thought you always had to have FSR below or = to 1 to work???

TCW did you try high settings?

 

I believe that FSR is scaling so much better that 100% (3168x3090) native render is significantly worse than 2668*2552 scaled high.  I was messing with 300% last night.  It might be my new sweet spot.  I'm going to make a new set of test stills in photoshop.

Edited September 6, 2021 by BH_Adabadoo_VR

[TCW_Brzi_Joe]

It looks great with high scale, but is it really better for spotting? It is hard to say, at least for me.

I tried up to scale 1.5-1.7, cant go higher with my 2080ti.

 

[BH_Adabadoo_VR]

1 hour ago, TCW_Brzi_Joe said:

It looks great with high scale, but is it really better for spotting? It is hard to say, at least for me.

I tried up to scale 1.5-1.7, cant go higher with my 2080ti.

 

For spotting I can't say.  I was hoping Fenris would see if the high scale looked better for him as well and then see if helped or hurt spotting.

I've had no issues with my spotting but I use different setting not nearly as well tested as written down here but I had different results with FSR which is now in spotting so I thought I'd share.  It may well be better image worse spotting as it will oversample the dot planes more??

[Ala13_UnopaUno_VR]

On 6/9/2021 at 1:23, SCG_Fenris_Wolf said:

- 06.09.2021 Actualización

- Incluido el Upscaler FidelityFX SuperResolution (FSR) de fholger y cómo usarlo.

- Configuración de SteamVR reelaborada para cada auricular a la luz de FSR

- Ajustado Vive Pro 2, Reverb G2 / 1, Samsung Odyssey, índice agregado y Vive Pro --- configuraciones dentro del juego

- Se volvió a agregar Pimax 8KX

I am not able to find your post with the update

[ITAF_Airone1989]

17 minutes ago, Ala13_UnopaUno_VR said:

I am not able to find your post with the update

He updated the first post of the topic

[Ala13_UnopaUno_VR]

this?

this???

[15[Span.]/JG51Costa]

Hola Unopauno.

No, simplemente ve a la primera página de este mismo hilo.

El primer post está actualizado.

 

Hello Unopauno.
No, just go to the first page of this thread.
The first post is updated.

 

Saludos.

15(Span.)/JG51 Costa.

Edited September 7, 2021 by 15[Span.]/JG51Costa

[TCW_Brzi_Joe]

It is an error in forum, two treads are fused together. Tread from Fenris start near bottom of page 6 of this mess.

link to tread from Fenris_Wolf

Edited September 7, 2021 by TCW_Brzi_Joe

[15[Span.]/JG51Costa]

You are right!
I hadn't even noticed it.
Yesterday when I read it, it was correct.
Wow, for once I'm trying to help... ?

 

Saludos.

15(Span.)/JG51 Costa 

[Ala13_UnopaUno_VR]

 

What can be caused by these spikes, while I'm flying at the finish?

[Giovanni_Giorgio]

55 minutes ago, Ala13_UnopaUno_VR said:

 

What can be caused by these spikes, while I'm flying at the finish?

 

Turning your head quickly to check six or 6/9 o'clock?