Jump to content

DCS VR Optimizations - MYTH BUSTED - A sort of peer review of the Lukas S 0.5 Pixel Density video


Recommended Posts

 

In this video I describe why the proposed 0.5 PD method for DCS shown by Lukas S does not actually work.  This is not a put down of the work Lukas has done for the community, he is a respected member of the community with lots of great optimization videos for VR and every one of you should go check his channel out.   I show how the PD Lukas used for his benchmarking was incorrect and how I came to the conclusion using RenderDoc and a Spreadsheet with the calculations which you can find below.

 

DCS SteamVR Resolution Calculator - https://docs.google.com/spreadsheets/d/1ygjuleAivvA4C-EnHU2Hc6AddoRzvBvpsxe8Wse8feE/edit?usp=sharing
Lukas S YT - https://www.youtube.com/channel/UC5QBBTAy6oXhzmb4zENxUQg
RenderDoc - https://renderdoc.org/
 

  • Like 5
  • Thanks 2

Twitch2DCS - Bring twitch chat into DCS.

SplashOneGaming.com - Splash One is a community built on combat flight simulation. S1G Discord

 

twitch / youtube / facebook / twitter / discord

Link to post
Share on other sites

Thanks Jabbers! a voice of sanity! 

 

The only thing i would suggest, is not talking about PD as percentage.. as i think that just adds to the confusion. great video!

 

also see a similar conclusion here 

 


Edited by speed-of-heat
  • Like 1

SYSTEM SPECS: Hardware Intel Corei7-9700K @ 5.1 GHz, 32Gb RAM, EVGA 3090 XC3, Dell S2716DG, Thrustmaster Warthog + MFG Crosswinds V2, HP Reverb G2 @ 2836 x 2768
SOFTWARE: Microsoft Windows 10 Pro x64, VoiceAttack & VIACOM PRO, TacView

VR Stuff:   My Settings for 3090 and G2  My Settings for 1080ti  Shaders MOD for VR  My variant of Kegetys mod with clear water and also IC PASS for current beta & stable, Patch Status

 

1569924735_WildcardsBadgerFAASig.jpg.dbb8c2a337e37c2bfb12855f86d70fd5.jpg

Link to post
Share on other sites

Much as I agree completely with what you're saying and have tested with correctly calculated relative resolutions on my rig and get absolutely no benefit at all, I think that if you watch his video carefully, in his defence, he states that he is aware that he is not comparing identical resolutions and what he is actually saying is that he believes that Steam Advanced Super sampling is better at generating a clearer (less jagged edged/less shimmering?) picture than DCS's, such that if you lower the workload in DCS and instead let Steam do the work you MAY get the same EXPERIENCE when the absolute comparative resolution is lower.

 

Doesn't work for me on my rig but he's suggesting that folk with systems that are struggling can lower the resolution in this way to gain FPS without losing so much PERCEIVED visual quality. I'm not saying it works, only that that is what Lukas is suggesting.

 

The problem is that this is not clear enough in the video and many people are taking it as a free FPS bump at the same resolution, which it clearly isn't.

RYZEN 5900X | 32GB | ASUS Strix RTX3090 | 500GB NVMe OS 1000GB NVMe DCS | Warthog HOTAS | HP Reverb G2 | VA & ViacomPRO

Link to post
Share on other sites

People are looking for the "Holy Grail one-size-fits-all" of DCS VR settings and the state of the technology just isn't there yet.  There is an old saying that "A person's perception is their own reality" and this certainly applies to VR.  We each have our own sensitivity to video presentation.  When you also have to account for face shape, eyesight acuity, computer hardware and software, VR becomes a VERY personalized endeavor.  The settings are easy enough to play with, so keep notes and try them out.  I founds settings that worked for me by using info from several sources including the Lukas video's as a starting point, but they were not the final settings I ended up with.  I really appreciate the effort the participants of this community put into sharing their experiences.  Thank you Jabber for providing the math.  Lukas S your video helped me as well....Thanks!

Link to post
Share on other sites
2 hours ago, Coxy_99 said:

Both are actually wrong heres what DCS in reality looks like in VR: 

 

Untitled.png

 

Hey, where can I get that Legoland map ?

  • Like 2

RYZEN 5900X | 32GB | ASUS Strix RTX3090 | 500GB NVMe OS 1000GB NVMe DCS | Warthog HOTAS | HP Reverb G2 | VA & ViacomPRO

Link to post
Share on other sites
32 minutes ago, Coxy_99 said:

 

Just use VR in DCS 😆

 

..is there another way to fly ?

 

Screen_210327_170629.png

Screen_210327_170351.png

  • Like 1

RYZEN 5900X | 32GB | ASUS Strix RTX3090 | 500GB NVMe OS 1000GB NVMe DCS | Warthog HOTAS | HP Reverb G2 | VA & ViacomPRO

Link to post
Share on other sites
3 hours ago, Coxy_99 said:

Both are actually wrong heres what DCS in reality looks like in VR: 

 

Untitled.png

U need to upgrade your headset 🤣🤣🤣

  • Haha 1

i7 10700@2.6Ghz / RTX 3080 / Reverb G2 / TrackIR / Virpil Wrbrd + Alpha Stick + CM2 throttle / Thrustmaster TRP Rudder

Link to post
Share on other sites
1 hour ago, Brainfreeze said:

U need to upgrade your headset 🤣🤣🤣

 

Lol yeah maybe that was with a google cardboard or something.

Don B

 

My VKB Gunfighter MK III Pro L Review

EVGA Z390 Dark MB | i9 9900k CPU @ 5.2 GHz | EVGA RTX 3090 FTW3 Ultra | 32 GB G Skill Trident Z 3200 MHz CL14 | Corsair H150i Pro Cooler | Samsung 970 Pro, 970 Evo 1TB m.2 Nvme |VKB GF MK III Pro L Ultimate Grip| Virpil CM3 Throttle |Reverb G2 | 

Link to post
Share on other sites

Hud's nice and sharp tho' 😆

  • Like 1

Intel Core i7 6700K @ 4.5GHz. Asus-Z170-PRO MB

CORSAIR H105 HYDRO CPU COOLER.

EVGA GTX 1080Ti FTW3 Elite.

16GB DDR4 2666MHZ HYPERX SAVAGE.

SAMSUNG M.2 SSD 128GB SM951 Boot Drive.

SAMSUNG SSD 500GB EVO Working Drive.

Windows 10 Professional

Link to post
Share on other sites

To those of you who think I missed the point Lukas was trying to make. I hope you read this...

I understand that Lukas was trying to make a point about the resolution ultimately being lower, but my point is that it works out the same no matter how you swing the numbers, if the effective PD is the same in both scenarios.

In this google drive you have 3 images which are frame buffer dumps of the right eye from RenderDoc. These images represent the resolution that DCS is rendering to before being sent to the compositor for VR.

 

The first image is labeled "140-0.5-500 rt eye". This means, Global 140% Resolution Per Eye in SteamVR, 0.5 Pixel Density in DCS and 500% Custom Resolution Multiplier in the SteamVR Per Application settings. I averaged ~74 FPS at this setting in the F-16 Caucasus Free Flight mission.

 

The second image is "140-1.118238994-100 rt eye". This means, Global 140% Resolution Per Eye in SteamVR, 1.118238994 Pixel Density in DCS and 100% Custom Resolution Multiplier in the SteamVR Per Application settings. This is the exact equivilant of the 1st images settings but using PD only to scale it rather than SteamVR. I averaged the same ~74 FPS at this setting in the F-16 Caucasus Free Flight mission.

 

The third image is "140-1.118238994-100 rt eye". This means, Global 140% Resolution Per Eye in SteamVR, 1.4 Pixel Density in DCS and 100% Custom Resolution Multiplier in the SteamVR Per Application settings. This is the equivilant of the settings used by Lukas in his video to demonstraight the clarity is roughly the same vs the 0.5PD solution. I averaged the same ~58 FPS at this setting in the F-16 Caucasus Free Flight mission.

 

The argument was that the clarity in the 140/0.5/500 is better than the clarity from a PD of 1.4 as showing in Lukas' video. If you layer these images on top of each other and simply look at small pieces of the frame together, the clarity in the 1.4PD image is easily higher, but because it is downsampled to your HMD resolution, you do not see it in the end result. However, the 1st 2 images mentioned show that the clarity is actually slightly worse (only noticable when zoomed way in) compared to settings of the 2nd image which would be the equivilant frame size using DCS's PD as the scaling factor instead of SteamVR.

 

Now, all of these images then get downsampled into the HMD itself, and I would interject that this process may actually result in different qualities of clarity based on the HMD subpixel matrix used (ie. pentile, striped, etc), the resolution of the HMD itself, and ultimately the size of the image being downsampled. Give 2 images the first being 25% larger than the target resolution, vs the second being 75% larger than the target image, there is an argument to be made that the 2nd image could give you a worse quality downsampled image in clarity given the amount of antialiasing that occurs because of its initial resolution.

 

I hope this clears things up a bit and I encourage you all to take a look at the Effective PD value in the calculator and try it to see if it gives you the same/better/or worse results given the assumption that you dont change Resolution Per eye and reset Customer Resolution Multiplier back to 100%

 

https://drive.google.com/folderview?id=1q6n6rV2qWLk9ByzjExAK3IBkyEpaGcIO

  • Like 3
  • Thanks 1

Twitch2DCS - Bring twitch chat into DCS.

SplashOneGaming.com - Splash One is a community built on combat flight simulation. S1G Discord

 

twitch / youtube / facebook / twitter / discord

Link to post
Share on other sites
8 hours ago, Jabbers_ said:

To those of you who think I missed the point Lukas was trying to make. I hope you read this...

I

 

..That was me Jabbers and I apologise. I do really appreciate you taking the time to clarify.

  • Like 1

RYZEN 5900X | 32GB | ASUS Strix RTX3090 | 500GB NVMe OS 1000GB NVMe DCS | Warthog HOTAS | HP Reverb G2 | VA & ViacomPRO

Link to post
Share on other sites
On 3/27/2021 at 11:23 PM, ShaunX said:

 

..That was me Jabbers and I apologise. I do really appreciate you taking the time to clarify.

It was more than just you, and its not a big deal.   It was a point I was trying to make but it was not clear so I am happy to expand on it 😉

  • Like 1

Twitch2DCS - Bring twitch chat into DCS.

SplashOneGaming.com - Splash One is a community built on combat flight simulation. S1G Discord

 

twitch / youtube / facebook / twitter / discord

Link to post
Share on other sites
On 3/27/2021 at 7:53 AM, Jabbers_ said:

 

In this video I describe why the proposed 0.5 PD method for DCS shown by Lukas S does not actually work.  This is not a put down of the work Lukas has done for the community, he is a respected member of the community with lots of great optimization videos for VR and every one of you should go check his channel out.   I show how the PD Lukas used for his benchmarking was incorrect and how I came to the conclusion using RenderDoc and a Spreadsheet with the calculations which you can find below.

 

DCS SteamVR Resolution Calculator - https://docs.google.com/spreadsheets/d/1ygjuleAivvA4C-EnHU2Hc6AddoRzvBvpsxe8Wse8feE/edit?usp=sharing
Lukas S YT - https://www.youtube.com/channel/UC5QBBTAy6oXhzmb4zENxUQg
RenderDoc - https://renderdoc.org/
 

Maybe now that Jabbers is saying what some of us have been banging their head against a brick wall about for a couple of years now means that some people will actually listen!

  • Like 3

Intel i9 10900K 5.2GHz · MSI Gaming X Trio RTX 3090 · ASUS ROG STRIX Z590-F Wi-Fi · Acer 4K 32" XB321HK · Samsung 970 500Gb M.2 NVMe · 2 x Samsung 850 Evo 1Tb · 2Tb HDD · 32Gb Corsair Vengance 3000MHz DDR4 · Windows 10 · Thrustmaster TPR Pedals · TrackIR5 · Thrustmaster F/A-18 Hornet Grip · Virpil WarBRD Base · Virpil Throttle MT-50 CM3 · Virpil Alpha Grip · HP Reverb G2

Link to post
Share on other sites
55 minutes ago, imacken said:

Maybe now that Jabbers is saying what some of us have been banging their head against a brick wall about for a couple of years now means that some people will actually listen!

nah.  🤣🤣🤯

  • Like 1

i7 10700@2.6Ghz / RTX 3080 / Reverb G2 / TrackIR / Virpil Wrbrd + Alpha Stick + CM2 throttle / Thrustmaster TRP Rudder

Link to post
Share on other sites

Sadly, you're correct! :sad_2:

  • Haha 1

Intel i9 10900K 5.2GHz · MSI Gaming X Trio RTX 3090 · ASUS ROG STRIX Z590-F Wi-Fi · Acer 4K 32" XB321HK · Samsung 970 500Gb M.2 NVMe · 2 x Samsung 850 Evo 1Tb · 2Tb HDD · 32Gb Corsair Vengance 3000MHz DDR4 · Windows 10 · Thrustmaster TPR Pedals · TrackIR5 · Thrustmaster F/A-18 Hornet Grip · Virpil WarBRD Base · Virpil Throttle MT-50 CM3 · Virpil Alpha Grip · HP Reverb G2

Link to post
Share on other sites

What I see is that comparing 1.0 PD / 100% and 0.5 PD / 500% on the G2 I get what appears to be a much worse quality (but faster) image on the 0.5/500%. Unless the scaling for distortion is a function of presented resolution (which I imagine it isn't?), more pixels should be rendered at 0.5PD/500% so this doesn't really make sense to me.

Link to post
Share on other sites
  • 2 weeks later...
On 3/31/2021 at 3:23 PM, tigger6 said:

What I see is that comparing 1.0 PD / 100% and 0.5 PD / 500% on the G2 I get what appears to be a much worse quality (but faster) image on the 0.5/500%. Unless the scaling for distortion is a function of presented resolution (which I imagine it isn't?), more pixels should be rendered at 0.5PD/500% so this doesn't really make sense to me.

This is highly dependent on your settings.   Do you for instance have MSAA turned on?  Super Sampling above the native resolution provides AA when downsampling to fit, combining this with MSAA can make things look worse.   There are so many variables in all of this, and that's a huge reason so many of the results are subjective.

Twitch2DCS - Bring twitch chat into DCS.

SplashOneGaming.com - Splash One is a community built on combat flight simulation. S1G Discord

 

twitch / youtube / facebook / twitter / discord

Link to post
Share on other sites
4 hours ago, Jabbers_ said:

This is highly dependent on your settings.   Do you for instance have MSAA turned on?  Super Sampling above the native resolution provides AA when downsampling to fit, combining this with MSAA can make things look worse.   There are so many variables in all of this, and that's a huge reason so many of the results are subjective.

Good point. I didn't think about other post processing. I do have MSAA turned on because I have pretty terrible jagginess on other airplanes sitting on the ramp in most cases. Not investing 100% effort into it, VR optimization seems to be not that easy...

Link to post
Share on other sites
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...