Another win for Fholger

[javelina1]

thanks SOH!  I'm gonna grab this one as well.

[EvilBert VR]

Hey everyone.

Has anyone noticed major changes with the latest 1.2 version of the shader?

TIA

[Sr.]

1.3 just released

https://github.com/fholger/openvr_fsr/releases

[Taz1004]

I just found this and while it works well, up sampling doesn't seem to get applied to the MFD fonts.  They become unreadable even at Ultra.  Is that same with everyone?

[speed-of-heat]

@Taz1004 my MFD's are certainly readable at ultra... with the exception of the unmodded AMPCD, which is meant to be that way 

 

[Taz1004]

7 minutes ago, speed-of-heat said:

@Taz1004 my MFD's are certainly readable at ultra... with the exception of the unmodded AMPCD, which is meant to be that way 

 

But it does get blurrier?  I'm trying to figure out if that's the way it is or I'm doing something wrong.

[speed-of-heat]

maybe a little ... but not by much ... certainly not enough for me to have asked the same question 

[dutchili]

I tried the previous version a while back but did not notice a difference. When i enabled the debug mode i did not see the visualization of the super sampling area.

I now noticed the GitHub text says 'steamvr'.

 

Should this work for Oculus VR too?

SteamVR is much slower for me than Oculus VR -> do i need to configure something, or is this the common perception with the Oculus Quest 2.

[speed-of-heat]

26 minutes ago, dutchili said:

I tried the previous version a while back but did not notice a difference. When i enabled the debug mode i did not see the visualization of the super sampling area.

I now noticed the GitHub text says 'steamvr'.

 

Should this work for Oculus VR too?

SteamVR is much slower for me than Oculus VR -> do i need to configure something, or is this the common perception with the Oculus Quest 2.

 

I don't believe it works directly with Oculus, but i don't have a Oculus headset or any experience, but that might be a better question for the mod author on their github.

[Taz1004]

26 minutes ago, dutchili said:

I tried the previous version a while back but did not notice a difference. When i enabled the debug mode i did not see the visualization of the super sampling area.

I now noticed the GitHub text says 'steamvr'.

 

Should this work for Oculus VR too?

SteamVR is much slower for me than Oculus VR -> do i need to configure something, or is this the common perception with the Oculus Quest 2.

Make sure radius is lower than 1 and debug visualization only shows up in HMD.  Not on the mirror.  If you still don't see it, I guess it's not working.  Says it's for SteamVR only.

[dutchili]

Confirmed, only works with stream VR which gives me less performance them oculus

[Sr.]

Version 2.0 just hit

 

https://github.com/fholger/openvr_fsr/releases

[speed-of-heat]

Thanks!

[Hartsblade]

47 minutes ago, Sr. said:

Version 2.0 just hit

 

https://github.com/fholger/openvr_fsr/releases

Thanks!

[Sr.]

First impression v2

After swapping back and forth several times, I think (for me anyway) NIS provides more crisp and readable MFD fonts in Hornet and Viper.

[Calinho]

37 minutes ago, Sr. said:

First impression v2

After swapping back and forth several times, I think (for me anyway) NIS provides more crisp and readable MFD fonts in Hornet and Viper.

yep, for me NIS is better for the cockpit, the rest is the same: terrain, performance. My dream for VR is viable anti aliasing option, maybe one day

Edited November 18, 2021 by Calinho

[Taz1004]

Thanks.  With NIS I can be much more aggressive without sacrificing MFD readability.  With FSR, I couldn't do less than 0.8.  With NIS, even 0.67 looks good.

[javelina1]

4 hours ago, Sr. said:

Version 2.0 just hit

 

https://github.com/fholger/openvr_fsr/releases

cool beans, thanks!

[DucS2r]

1 hour ago, Taz1004 said:

Thanks.  With NIS I can be much more aggressive without sacrificing MFD readability.  With FSR, I couldn't do less than 0.8.  With NIS, even 0.67 looks good.

For render scale? 

[Taz1004]

29 minutes ago, DucS2r said:

For render scale? 

Yes

[JayRoc]

I use AMD FSR as my antialaising which works pretty well in DCS, together with ReShade i get a crisp image.

Steam SS 100%
Steam per App SS 65%

    "enabled": true,
    "renderScale": 1.18,
    "sharpness": 0.9,
    "radius": 0.75,
    "applyMIPBias": true,
    "debugMode": false

Here is FSR vs NIS for this case. But the antialaising works better with AMD FSR - means with this settings FSR looks better for me.

AMD FSR with Reshade 12.8ms

AMD FSR without ReShade 12.2ms

Nvidia NIS with Reshade 13.0ms

Nvidia NIS without ReShade 12.5ms

Testet with 21.40 beta driver. I feel like the sideways head movement works more fluent with FSR? But might be me.

[JayRoc]

Steam SS 100%
Steam per App SS 100%

    "enabled": true,
    "renderScale": 0.68,
    "sharpness": 0.9,
    "radius": 0.75,
    "applyMIPBias": true,
    "debugMode": false

AMD FSR without ReShade

Nvidia NIS without Reshade

The instruments look more crisp with Nvidia NIS but the distant, clouds and flimmering/genral qualitiy looks better to me in FSR.
AMD FSR seems to be slightly faster on AMD RX6800XT, but its so marginal that my decision would base on the quality you get for your setup

[speed-of-heat]

on like for like testing i got:

			Frametime in ms		Delta From Baseline
Ver	Test	Baseline	GPU	CPU	GPU	CPU	Changes
2.7.7.15038	104	Test 100	10.72	6.25			Alderlake
2.7.8.16140	107	Test 104	10.50	6.19	0.22	0.06	Initial testing 2.7.8 FSR
2.7.8.16140	108	Test 104	10.49	6.12	0.23	0.13	NIS instead of FSR

functionally runtime variance between the two in terms of performance 

[Sile]

With NIS the sharpening seems to be more pronounced than with FSR, when the same "sharpness" config value is compared.
Maybe that's the reason why some people notice more flimmering / better cockpit readability with NIS.

With NIS the sharpness can be set about 0.2 to 0.3 lower than with FSR vor the same effect.

 

[Taz1004]

37 minutes ago, Sile said:

With NIS the sharpening seems to be more pronounced than with FSR, when the same "sharpness" config value is compared.
Maybe that's the reason why some people notice more flimmering / better cockpit readability with NIS.

With NIS the sharpness can be set about 0.2 to 0.3 lower than with FSR vor the same effect.

It may depend on the headset and you may not notice it with higher res headsets.  But for Vive Pro, F18 MFD was unreadable with anything less than FSR 0.8 regardless of sharpness setting.  But all other cockpit button labels looked fine.  So it looked like upsampling wasn't being applied to MFD with FSR.  So simple fact that I can use lower render scale with NIS yields higher performance benefit.