Varjo Aero: Общее руководство для новых владельцев

[Supmua]

2 hours ago, Yuma said:

I did the full repair (File integrity in Steam) and nothing. I unistalled DCS and re-installed it, still have the issue. Going over to Discord to see if I can find a solution and then Varjo, I guess.

Yeah see if there are others with similar issues.

[Supmua]

7 hours ago, Gordy said:

I am not seeing the foveated rendering "box" artifact, even in hornet nevada free flight looking straight up... but I am running with default brightness and gamma levels, so maybe that's a factor.

I *do* absolutely see a weirdly rendered box if I turn on the night vision in the Mi-24... like one side of a set of binoculars or goggles are floating in front of you, and it follows your eye tracking.

I tried turning up the focus_multiplier to 1.2 and 1.5, and it looks marginally better than 1.0 without a dramatic loss of FPS - I still run at 90 fps most of the time, but can come down to 75-80 fps on landings or crowded areas like around Las vegas.    I mostly feel like shimmering is less, distance vision is better, and fine details in the cockpit are nicer, I'll probably run with this set to 1.2 unless the performance hit turns out to be too much at critical times.   I'm hoping this will improve spotting bandits in the distance. 

I need to look up what the max resolution of the focus area on the aero is again 

Also... this new functionality means a VR-3 or XR-3 can now take advantage of their discrete focus displays, yes?   (not that I need to go that far)

I also have performance drop with focus at 1.5x, for regular flying it’s fine but for serious MP or campaign mission I’d take performance over visuals any day so it’s back to 1.0 or even 0.9–still haven’t decided what’s best for my current setup.

Not sure about the non-Aero headsets but I think they use the same framework for DFR, meaning they should have even better visual especially with that extra OLED screen.  Dunno about performance though.

EDIT:  It's also interesting that for the first time I can truly appreciate the difference between textures at high vs medium in cockpit setting, via supersampling of the focus screen in DFR mode.  On the F/A-18C's UFC, the white colored numbers and letters on the keypads are slightly blurry on medium textures compared to crisp LCD and MFD texts.

Edited March 27, 2023 by Supmua

[Supmua]

@YumaOne thing I can think of that may help is to run fpsVR before starting DCS (which means you have to run SteamVR GUI first).  This is to cross check whether you are actually using Varjo OpenXR or not, because this runtime is not compatible with fpsVR which only runs via Steam layer (kinda like Reshade 5). 

If you can see fpsVR popup screen in DCS, this means you're either running OpenVR or Steam OpenXR (fpsVR supports both Steam OpenXR and Steam OpenVR), rather than Varjo OpenXR runtime.  It might be that launching DCS.exe directly with Steam version may not work as intended.  So if you don't see fpsVR popup window in DCS then you're golden.

Also if you run DCS via OpenVR runtime, OpenXR Explorer doesn't know this so it will default to showing Varjo's desktop setting (which is the most current active OpenXR runtime) which may give false impression that you are running DCS correctly.  If you use Steam OpenXR to run DCS, then it will show up as SteamVR/OpenXR in OpenXR Explorer.

Edited March 27, 2023 by Supmua

[Gordy]

3 hours ago, Supmua said:

I also have performance drop with focus at 1.5x, for regular flying it’s fine but for serious MP or campaign mission I’d take performance over visuals any day so it’s back to 1.0 or even 0.9–still haven’t decided what’s best for my current setup.

Not sure about the non-Aero headsets but I think they use the same framework for DFR, meaning they should have even better visual especially with that extra OLED screen.  Dunno about performance though.

EDIT:  It's also interesting that for the first time I can truly appreciate the difference between textures at high vs medium in cockpit setting, via supersampling of the focus screen in DFR mode.  On the F/A-18C's UFC, the white colored numbers and letters on the keypads are slightly blurry on medium textures compared to crisp LCD and MFD texts.

 

I thought someone said that there were issues (crashing?) at multipliers of < 1.0, so I'd try reducing PPD in varjo base, then dial the multipliers up to where you want them instead.  Presumably you'd get the result you want that way.

Also, yes... I noticed that some of the gauges have recessed painted numbers now that I'm running with finer resolution; I'd never noticed that before.   Also, I lean forward less to read labels, MFDs, and gauges, because they're crisp even when sitting up straight.   Also, the houses, trees, buildings, etc., all look much more real and clean from altitude or during landing approaches. 

What I *really* want to see is whether I can spot those darned gazelles at long distances on blueflag, now that my distance vision appears to be cleaner...

[Supmua]

14 minutes ago, Gordy said:

I thought someone said that there were issues (crashing?) at multipliers of < 1.0, so I'd try reducing PPD in varjo base, then dial the multipliers up to where you want them instead.  Presumably you'd get the result you want that way.

Also, yes... I noticed that some of the gauges have recessed painted numbers now that I'm running with finer resolution; I'd never noticed that before.   Also, I lean forward less to read labels, MFDs, and gauges, because they're crisp even when sitting up straight.   Also, the houses, trees, buildings, etc., all look much more real and clean from altitude or during landing approaches. 

What I *really* want to see is whether I can spot those darned gazelles at long distances on blueflag, now that my distance vision appears to be cleaner...

I crashed with peripheral mult <1 at 37 and 35 PPD (not at 39 PPD).  No issue so far with focus mult <1.  Despite increase in clarity, I still find that ground spotting without TPOD remains difficult, still have to rely on VR zoom at times.

[Yuma]

4 hours ago, Supmua said:

Yeah see if there are others with similar issues.

Got it to work! Don't ask me how but I had to get the stanalone version, open beta instead of Steam's and it's now working flawlessly. I definatlely and strongly suggest standalone version of DCS. No more shaking nor any double cursor crap and better performance, at least for me.

[Supmua]

6 hours ago, Yuma said:

Got it to work! Don't ask me how but I had to get the stanalone version, open beta instead of Steam's and it's now working flawlessly. I definatlely and strongly suggest standalone version of DCS. No more shaking nor any double cursor crap and better performance, at least for me.

So stand alone works for you but not the Steam version.  Interesting.  I don’t use the Steam version nor have any more storage to install it so won’t be able to check this out.  I think most DCS users/old timers here also use the stand alone version. 
 

If you want to look further though, run fpsVR then run your Steam version of DCS and see if fpsVR works or not. 

Edited March 27, 2023 by Supmua

[Gordy]

Tested this a little bit at the Highest PPD setting - landing F18 nevada free flight at McCallan, my fps goes from:

SSAA off, MFAA off, focus_multiplier=2.0:  90 fps in flight to 69-70 fps on the tarmac.   Some stuttering on final. 
SSAA 1.5, MFAA off, focus_multiplier=2.0:  90 fps in flight to 69-70 fps on the tarmac.  (no degradation from turning on SSAA)

SSAA 1.5, MFAA off, focus_multiplier=1.0:   90 fps in flight to 80 fps on the tarmac.
SSAA off, MFAA off, focus_multiplier=1.0:   90 fps in flight to 77-78 fps on the tarmac.   (huh, SSAA 1.5 performed better)
SSAA 2.0, MFAA off, focus_multiplier=1.0:  90 fps in flight to 79-80 fps on the tarmac.
SSAA off, MFAA 2x, focus_multiplier=1.0:  90 fps in flight to 72 fps on the tarmac. 

SSAA off, MFAA 2x, focus_multiplier=1.5: 90 fps in flight to 69 fps on the tarmac
SSAA off, MFAA off, focus_multiplier=1.5: 90 fps in flight to 73-74 fps on the tarmac
SSAA 2.0, MFAA off, focus_multiplier=1.5: 90 fps in flight to 73-76 fps on the tarmac

noticeable 'softening' of cockpit and distance vision going from  focus_multiplier=2.0 to 1.0

You probably won't notice much improvement inside the cockpit above 1.2. 

1.5 looks really good, I think because it's fine enough that a lot of the computer-graphic-y stuff isn't in your face - I see a lot less shimmer, etc., and buildings in the distance just look like buildings, and houses, cars, etc far below me look really good.

2.0 looks amazing - but the only big step up over 1.5 is that I think I can see more clearly at 20nm and beyond.  Looking beyond vegas while approaching mcCallan, instead of a tan smudgefield at that distance, I swear I can see detail that is glossed over at 1.5.   I have yet to see if this helps spot traffic, I'll have to try replaying some old track files to compare. 

When you get close enough to certain objects (look at the vegas strip from the runway), there wasn't much difference, there's always some aliasing on the lines of the pyramid, for example. 

MFAA is clearly more expensive than SSAA for me, contrary to the tooltips in the dcs options menu. 
Toggling off the mirrors in the F18 took me from 72 to 77 fps, so that might be a useful bump on landings or if you're doing anything demanding but don't need the mirrors. 

[Yuma]

Where is that setting? focus_multiplier located? Inside Varjo folder settings?

6 hours ago, Supmua said:

So stand alone works for you but not the Steam version.  Interesting.  I don’t use the Steam version nor have any more storage to install it so won’t be able to check this out.  I think most DCS users/old timers here also use the stand alone version. 
 

If you want to look further though, run fpsVR then run your Steam version of DCS and see if fpsVR works or not. 

 

I learned that on the main menu in DCS at the bottom right in white letters it tells you Multi Thread Version, i did't notice that in Steam. I will check it later as I still have it installed. What tool/software do you use for checking FPS in Varjo OpenXR thou?

[Gordy]

@Yuma see https://github.com/mbucchia/Varjo-Foveated/wiki

[Supmua]

6 hours ago, Yuma said:

Where is that setting? focus_multiplier located? Inside Varjo folder settings?

I learned that on the main menu in DCS at the bottom right in white letters it tells you Multi Thread Version, i did't notice that in Steam. I will check it later as I still have it installed. What tool/software do you use for checking FPS in Varjo OpenXR thou?

Right now DCS fps counter works (RCTRL+PAUSE) with DCS mt via Varjo OpenXR DFR.  It shows fps and GPU frametimes, not ideal but better than nothing.  You can also use Varjo analytics tool but it’s cumbersome.

I've been observing GPU frametimes between Varjo OpenXR DFR vs Steam OpenXR and the difference is actually not that big (using DCS fps counter, GPU frametime tracing is the white line on the lower graph area).  GPU frametime in Varjo mode is around 1.0-1.5ms less (at 1.0 multiplier for both focus and peripheral screens), which is similar to what you can get with fixed foveated rendering in stereo view, although you can probably increase this gain by reducing the multiplier values.  Unfortunately there's no good way for me to check CPU frametimes in Varjo DFR mode, with Steam OpenXR it's pretty easy to see everything via fpsVR.

An ideal testing scenario would be to look at both CPU and GPU frametimes for multi-threaded Varjo OpenXR DFR vs nonDFR vs Steam OpenXR vs single-threaded counterparts.

Edited March 28, 2023 by Supmua

[Supmua]

Just reran the first mission from Op Inherence Resolve Campaign via Steam OpenXR and got some interesting findings.

Cons:

1. Visuals are not as crisp as DFR mode.

2. Max fps is not as good due to 1-1.5 ms higher GPU FT.  (This is mitigated by reducing SteamVR Sampling to 90%)

Pros:

1. No 1-1.5 sec pauses during the entire mission (these pauses were seen occasionally with DFR mode).

2. Min fps is much better.  Taxiing to the runway from parking I never dropped below 60 fps, whereas I saw as low as 40s with Varjo OpenXR DFR.

3. No white square box in the center of view.  Once you've seen this you can't unsee it even though how faint it might be.

 

So Steam OpenXR actually gave me smoother overall gameplay without any hiccups even though visually it's not as good.  Once I'm airborne I'm also 90 fps in this mode so the performance is up there with the right settings.  Note that I haven't changed any DCS settings, just downsampling SteamVR to 90% to equalize GPU frametimes.

Edited March 28, 2023 by Supmua

[Slick Twitchy]

OMG (yes I said that)-set focus to 1.5.  Peripheral to .7 in (C:\Program Files\OpenXR-Varjo-Foveated); So clear and crisp- I can make out ground units from 7,000 feet. Missile trails from miles away, no jaggies on wingman; cockpit info is super crisp.  Smooth (as far as I can tell)- I stopped using FPS counters so super subjective. Not super informative I know.  But just ecstatic about these new results and updates. 

[motoadve]

Is increasing focus to a value above 1 the same as increasing pixel density  in the VR tab in game?

What we are doing is increasing resolution, or is there a difference?

[Supmua]

2 hours ago, motoadve said:

Is increasing focus to a value above 1 the same as increasing pixel density  in the VR tab in game?

What we are doing is increasing resolution, or is there a difference?

It's the same concept but not the exact same because with pixel density setting you change the resolution of the entire screen, whereas with DFR mode you have more flexibility such as downsampling the peripheral screen while upsampling the focus screen.

You can also see the exact rendered resolution after PD setting changes using OpenXR explorer or OXRTK log file, at least with the non-DFR mode.  I haven't yet had a chance to checked with DFR mode (quad view) which works differently than stereo view.

Edited March 28, 2023 by Supmua

[Supmua]

10 hours ago, Slick Twitchy said:

OMG (yes I said that)-set focus to 1.5.  Peripheral to .7 in (C:\Program Files\OpenXR-Varjo-Foveated); So clear and crisp- I can make out ground units from 7,000 feet. Missile trails from miles away, no jaggies on wingman; cockpit info is super crisp.  Smooth (as far as I can tell)- I stopped using FPS counters so super subjective. Not super informative I know.  But just ecstatic about these new results and updates. 

Yeah focus 1.5x is something else in terms of clarity.

[motoadve]

Ok, so PD in game to 1.2 increases the resolution of the whole screen, then you add DFR , focus 1.0 (no changes) Pheripheral 0.7 lowers the res on the pheripheral area from 1.2= 84%

DFR Focus to 1.2 increases the resolution where your eyes look, 0.7 pheripheral reduces the resolution to the pheripheral area from 1.0 so should have a better performance moving the values in DFR =70%

Now I read somewhere the PD in game every.1 you add is like 20% resolution?
Is there a file where we can see the resolution I am playing at? (I dont use OXR tool kit) so I have the same resolution as with 1.2 PD in game ?

need to know what value I should change focus to.

[darkman222]

11 minutes ago, motoadve said:

Is there a file where we can see the resolution I am playing at? (I dont use OXR tool kit)

Exacly my question. Its not I dont use OXRTK. I just cant with DFR. So I will trust what Varjo Base tells me and from my previous experiences with fpsVR which kept telling me the high VB setting is 4148*3556. Would be cool to confirm this with a third party software like OXRTK or fpsVR as well as CPU frame times for comparison.

[Supmua]

I'm at work so can't do any testing, but try OpenXR Explorer tool (https://github.com/maluoi/openxr-explorer/releases) and see what it shows.

[dburne]

3 hours ago, motoadve said:

Ok, so PD in game to 1.2 increases the resolution of the whole screen, then you add DFR , focus 1.0 (no changes) Pheripheral 0.7 lowers the res on the pheripheral area from 1.2= 84%

DFR Focus to 1.2 increases the resolution where your eyes look, 0.7 pheripheral reduces the resolution to the pheripheral area from 1.0 so should have a better performance moving the values in DFR =70%

Now I read somewhere the PD in game every.1 you add is like 20% resolution?
Is there a file where we can see the resolution I am playing at? (I dont use OXR tool kit) so I have the same resolution as with 1.2 PD in game ?

need to know what value I should change focus to.

 

Guys leave PD in DCS at 1. Just work with the focus and peripherary resolutions in Varjo Foveated. By default they are each at 1. Try increasing focus whilst decreasing periphery to keep performance. 

Start out with your Varjo Base resolution you wish to use ( I use very high). And then work the DFR from there. Really no reason to throw DCS PD into the mix, let the Varjo software handle it.

Edited March 28, 2023 by dburne

[dburne]

Well I tried changing the numbers for focus and peripheral but ended up wtih some instability - DCS crashing to desktop upon trying to launch or freezing up when launching and having to close with task manager. Went back to original settings and all is well again - reckon I will leave it there. Still looks great and the performance gain was quite nice.

[Supmua]

43 minutes ago, dburne said:

Well I tried changing the numbers for focus and peripheral but ended up wtih some instability - DCS crashing to desktop upon trying to launch or freezing up when launching and having to close with task manager. Went back to original settings and all is well again - reckon I will leave it there. Still looks great and the performance gain was quite nice.

This is why some of us here lower the PD setting instead of peripheral_multiplier.  Focus_multiplier is pretty safe, but reducing peripheral_multiplier can cause crashing issue unless you're using 39 PPD.

[motoadve]

39PPD with motion smoothing any value below 0.7 for peripheral DCS crashes

Without motion smoothing, you can put any value for peripheral and no crashes.

Currently running 1.2 Focus, and 0.7 in peripheral, would be ideal if we could use a lower value here ,so Motion Smoothing can take advantage of some performance..

[Supmua]

18 minutes ago, motoadve said:

39PPD with motion smoothing any value below 0.7 for peripheral DCS crashes

Without motion smoothing, you can put any value for peripheral and no crashes.

Currently running 1.2 Focus, and 0.7 in peripheral, would be ideal if we could use a lower value here ,so Motion Smoothing can take advantage of some performance..

Good to know.  I've never used values below 0.7.

I'm actually seeing much more stable performance in Steam OpenXR than the DFR mode in complex missions.  For example in the Hornet Syria DLC campaign 1st mission, there would be intermittent GPU FT spike to >22ms on my setup while taxiing to the runway after the 2 Mirage jets.  Compared to Steam OpenXR run, my GPU FT never exceeds 13ms with zero spikes (13 vs 22 is huge by any standard), and this is reproducible since I've run this mission at least 4-5 times now.  Steam OpenXR also shows no hiccups through out the entire mission, while Varjo DFR can cause 1-1.5 sec hiccups on occasion.  I'm not saying this will happen with everyone, but on my setup Steam OpenXR provides smoother performance even though not visually as pleasing.  I have to try Varjo OpenXR stereo mode to see if this mirrors the Steam method.

Edited March 28, 2023 by Supmua

[dburne]

1 hour ago, motoadve said:

39PPD with motion smoothing any value below 0.7 for peripheral DCS crashes

Without motion smoothing, you can put any value for peripheral and no crashes.

Currently running 1.2 Focus, and 0.7 in peripheral, would be ideal if we could use a lower value here ,so Motion Smoothing can take advantage of some performance..

 

Interesting I was using Motion Smoothing and 37 PPD. May try it again tomorrow (reducing periphery) without Motion Smoothing and see how it does.