WipeUout Posted January 16, 2022 Posted January 16, 2022 After seeing a guy on youtube doing VR with SSAA, I decided to do a bit of testing to see the gain in quality compared with MSAA, which I've been using. I always though that SSAA would have a bigger demand on resources compared to MSAA as mention in the text bubbles in settings. Seems it is not quite the result I expected. Let's start with the settings I used: Then, I re-played @speed-of-heat's track (https://1drv.ms/u/s!Ar2cEt5yMycgm6skZyUPKfQMjm4mKw?e=so7zI5) using fpsVR to check my FPS average. Average FPS with MSAA off and SSAA 2x: 66.5 Average FPS with SSAA off and MSAA 2x: 56.2 Not only is SSAA much more efficient, the quality of the image looks way sharper!!! I am still wondering if this 18% improvement in FPS is real or I am dreaming... and it is such a nicer image! BTW I also supersample by 150% in steam VR, my Pimax 5kS is at standard FOV/90hz/1.0 rendering and PD set to 1.0 in DCS also, no image smoothing whatsoever. Anyone having similar results? Maybe this is very hardware specific. ------------------------------------------------------------------------------------------------------------------------------------------------------------ 9800X3D, RTX 4090, 96GB DDR 5, MSI Tomahawk 870E, Crucial 2TB x 2, TM WARTHOG COMBO + PENDULAR RUDDER PEDALS, THE AMAZING PIMAX 8K X, Sony 5.1 Spks+SubW | DCS, A-10C_II, AH-64D, F-14/15E/16/18, F-86F, AV-8B, M-2000C, SA342, Huey, Spitfire, FC3.
speed-of-heat Posted January 16, 2022 Posted January 16, 2022 so it depends on why you use MSAA, if you use it just for smoothing then its fine and you can swap over to SSAA, but if you use it to reduce shimmer, then SSAA doesnt help at all SYSTEM SPECS: Hardware AMD 9800X3D, 64Gb RAM, 4090 FE, Virpil T50CM3 Throttle, WinWIng Orion 2 & F-16EX + MFG Crosswinds V2, Varjo Aero SOFTWARE: Microsoft Windows 11, VoiceAttack & VAICOM PRO YOUTUBE CHANNEL: @speed-of-heat
Werewolf_fs Posted January 16, 2022 Posted January 16, 2022 (edited) hace 5 horas, WipeUout dijo: Not only is SSAA much more efficient, the quality of the image looks way sharper!!! SSAA is not applicable in VR Edited January 16, 2022 by Werewolf_fs 1
WipeUout Posted January 17, 2022 Author Posted January 17, 2022 I repeated the test several times (6-7) and results for each setting are max 0.2 FPS apart. Here are the averages: MSAA(2x) On: 56.2 SSAA(2x) On: 66.5 MSAA/SSAA Off: 67.3 MSAA alone has a 19.8% impact on my FPS, SSAA alone has a 1.2% impact only. I do see a small change using SSAA, could it be HMD dependent? SSAA does improve slightly the image quality but not much and leaves a lot of shimmering. It is MSAA that makes things more blurry in my VR headset with a high FPS drop. What would you suggest as a good compromise to reduce shimmering? Use a reshader? ------------------------------------------------------------------------------------------------------------------------------------------------------------ 9800X3D, RTX 4090, 96GB DDR 5, MSI Tomahawk 870E, Crucial 2TB x 2, TM WARTHOG COMBO + PENDULAR RUDDER PEDALS, THE AMAZING PIMAX 8K X, Sony 5.1 Spks+SubW | DCS, A-10C_II, AH-64D, F-14/15E/16/18, F-86F, AV-8B, M-2000C, SA342, Huey, Spitfire, FC3.
speed-of-heat Posted January 17, 2022 Posted January 17, 2022 1.2% is well inside margin of error... SYSTEM SPECS: Hardware AMD 9800X3D, 64Gb RAM, 4090 FE, Virpil T50CM3 Throttle, WinWIng Orion 2 & F-16EX + MFG Crosswinds V2, Varjo Aero SOFTWARE: Microsoft Windows 11, VoiceAttack & VAICOM PRO YOUTUBE CHANNEL: @speed-of-heat
WipeUout Posted January 17, 2022 Author Posted January 17, 2022 May I ask how do you define (or calculate) the margin of error? When a test is performed several times, the "margin of error" should be the range of results you get for the same conditions, or the difference between the lowest result and the highest result. In my test, I get a range of result of 0.2 between the highest and the lowest result. The FPS is pretty consistent. ------------------------------------------------------------------------------------------------------------------------------------------------------------ 9800X3D, RTX 4090, 96GB DDR 5, MSI Tomahawk 870E, Crucial 2TB x 2, TM WARTHOG COMBO + PENDULAR RUDDER PEDALS, THE AMAZING PIMAX 8K X, Sony 5.1 Spks+SubW | DCS, A-10C_II, AH-64D, F-14/15E/16/18, F-86F, AV-8B, M-2000C, SA342, Huey, Spitfire, FC3.
speed-of-heat Posted January 17, 2022 Posted January 17, 2022 Honestly much under 5% is likely to be run to run variance. net net if it’s not having an impact in cpu or gpu times it’s likely not actually doing anything… e.g. you don’t get anything for free. YMMV SYSTEM SPECS: Hardware AMD 9800X3D, 64Gb RAM, 4090 FE, Virpil T50CM3 Throttle, WinWIng Orion 2 & F-16EX + MFG Crosswinds V2, Varjo Aero SOFTWARE: Microsoft Windows 11, VoiceAttack & VAICOM PRO YOUTUBE CHANNEL: @speed-of-heat
edmuss Posted January 17, 2022 Posted January 17, 2022 I found MSAA 2x with MFAA enabled in nvcpl gives a good compromise of shimmer reduction and performance. 2x MSAA on its own doesn't do a great deal for the shimmer but the addition of MFAA is almost the equivalent of 4x MSAA but without the massive FPS hit. Shimmer is still there but it's greatly reduced, using a flyover of Beirut with lots of straight eedge is the worst areas for me. I dare say you could reduce the PD and run MSAA+MFAA and get equal image clarity but with more performance. I did run a couple of tests of SSAA and it didn't seem any better or worse really than MSAA. Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat Shoehorned into a 43" x 43" cupboard.
speed-of-heat Posted January 17, 2022 Posted January 17, 2022 there is a very helpful summary in the pinned comment. SYSTEM SPECS: Hardware AMD 9800X3D, 64Gb RAM, 4090 FE, Virpil T50CM3 Throttle, WinWIng Orion 2 & F-16EX + MFG Crosswinds V2, Varjo Aero SOFTWARE: Microsoft Windows 11, VoiceAttack & VAICOM PRO YOUTUBE CHANNEL: @speed-of-heat
dburne Posted January 18, 2022 Posted January 18, 2022 3 hours ago, edmuss said: I found MSAA 2x with MFAA enabled in nvcpl gives a good compromise of shimmer reduction and performance. 2x MSAA on its own doesn't do a great deal for the shimmer but the addition of MFAA is almost the equivalent of 4x MSAA but without the massive FPS hit. Shimmer is still there but it's greatly reduced, using a flyover of Beirut with lots of straight eedge is the worst areas for me. I dare say you could reduce the PD and run MSAA+MFAA and get equal image clarity but with more performance. I did run a couple of tests of SSAA and it didn't seem any better or worse really than MSAA. Yeah I have been running MFAA with my 2x MSAA for some time now. 1 Don B EVGA Z390 Dark MB | i9 9900k CPU @ 5.1 GHz | Gigabyte 4090 OC | 64 GB Corsair Vengeance 3200 MHz CL16 | Corsair H150i Pro Cooler |Virpil CM3 Stick w/ Alpha Prime Grip 200mm ext| Virpil CM3 Throttle | VPC Rotor TCS Base w/ Alpha-L Grip| Point Control V2|Varjo Aero|
WipeUout Posted January 18, 2022 Author Posted January 18, 2022 16 hours ago, edmuss said: I found MSAA 2x with MFAA enabled in nvcpl gives a good compromise of shimmer reduction and performance. 2x MSAA on its own doesn't do a great deal for the shimmer but the addition of MFAA is almost the equivalent of 4x MSAA but without the massive FPS hit. Shimmer is still there but it's greatly reduced, using a flyover of Beirut with lots of straight eedge is the worst areas for me. I dare say you could reduce the PD and run MSAA+MFAA and get equal image clarity but with more performance. I did run a couple of tests of SSAA and it didn't seem any better or worse really than MSAA. I have an AMD card, no MFAA available unfortunately and AMD driver options do not seem to affect anything in VR... I will try your suggestion with lowering the PD, and also the steamVR SS back to 100%. ------------------------------------------------------------------------------------------------------------------------------------------------------------ 9800X3D, RTX 4090, 96GB DDR 5, MSI Tomahawk 870E, Crucial 2TB x 2, TM WARTHOG COMBO + PENDULAR RUDDER PEDALS, THE AMAZING PIMAX 8K X, Sony 5.1 Spks+SubW | DCS, A-10C_II, AH-64D, F-14/15E/16/18, F-86F, AV-8B, M-2000C, SA342, Huey, Spitfire, FC3.
Hiob Posted January 18, 2022 Posted January 18, 2022 To test if MSAA, SSAA, MFAA or any other AA is actually doing something, I recommend spawning on the Supercarrier and pointing the camera to the isle (take screenshots for comparison). With the amount of thin lines and edges cramped into a confined area it is pretty easy to spot improvements (or the lack of). MSAA is pretty taxing, SSAA even more. So if you don't get a FPS penalty for using SSAA it's most likely not working. "Muß ich denn jedes Mal, wenn ich sauge oder saugblase den Schlauchstecker in die Schlauchnut schieben?"
WipeUout Posted January 19, 2022 Author Posted January 19, 2022 Tried many combination and the best visual result is using MSAA with a reshader (https://github.com/fholger/reshade/releases/tag/openvr_alpha2). With the reshader, scaling down a bit my quality setting which do not have a huge impact visually (visibility range, anisotropic filter, cockpit resolution, forest details) and using motion re-projection, I can achieve very good quality and keep a smooth and steady 90 FPS with bit of ghosting when low and looking sideways... I was hoping to avoid using motion re-projection because of ghosting effect but since my HMD minimum refresh rate is 90Hz, it is possible to maintain 90 FPS in single player missions but definitely not in multiplayer. ------------------------------------------------------------------------------------------------------------------------------------------------------------ 9800X3D, RTX 4090, 96GB DDR 5, MSI Tomahawk 870E, Crucial 2TB x 2, TM WARTHOG COMBO + PENDULAR RUDDER PEDALS, THE AMAZING PIMAX 8K X, Sony 5.1 Spks+SubW | DCS, A-10C_II, AH-64D, F-14/15E/16/18, F-86F, AV-8B, M-2000C, SA342, Huey, Spitfire, FC3.
Hiob Posted January 19, 2022 Posted January 19, 2022 On a side note - did they change anything to steam vr? How do I force reprojection off? "Muß ich denn jedes Mal, wenn ich sauge oder saugblase den Schlauchstecker in die Schlauchnut schieben?"
edmuss Posted January 19, 2022 Posted January 19, 2022 Disable it in the wmr for steamvr menu, the normal steamvr motion smoothing options don't affect wmr. Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat Shoehorned into a 43" x 43" cupboard.
Hiob Posted January 19, 2022 Posted January 19, 2022 2 hours ago, edmuss said: Disable it in the wmr for steamvr menu, the normal steamvr motion smoothing options don't affect wmr. I'm using an Index - so no wmr... "Muß ich denn jedes Mal, wenn ich sauge oder saugblase den Schlauchstecker in die Schlauchnut schieben?"
edmuss Posted January 19, 2022 Posted January 19, 2022 Doh! Miles away, as you were Ryzen7 7800X3D / RTX3080ti / 64GB DDR5 4800 / Varjo Aero / Leap Motion / Kinect Headtracking TM 28" Warthog Deltasim Hotas / DIY Pendular Rudders / DIY Cyclic Maglock Trimmer / DIY Abris / TM TX 599 evo wheel / TM T3PA pro / DIY 7+1+Sequential Shifter / DIY Handbrake / Cobra Clubman Seat Shoehorned into a 43" x 43" cupboard.
Recommended Posts