fixed 2.9.6 quad views foveated Varjo Aero issue under 90fps

[Phantom711]

So despite the title of this thread, I guess anyone using QVFR is affected, no matter which headset?

I am asking for a friend... ("He" has a Quest Pro)

Jokes aside...I also have some weird flickering...

[Bones308th]

Guys, thanks so much for your replies. I’ll refer to this post and apply the recommendations when I get home if I have any issues. Thanks again. 

Ray

[zildac]

8 minutes ago, Phantom711 said:

So despite the title of this thread, I guess anyone using QVFR is affected, no matter which headset?

I am asking for a friend... ("He" has a Quest Pro)

Jokes aside...I also have some weird flickering...

Correct, unfortunately

[USA_Recon]

I just find it incomprehensible that you have this entire population of DCS community customers using this technology and somehow you release this and break the experience for all these folks ?   

Fix your testing process.   Wake up that your customers use eye tracking headsets.    

It's like a web company not testing if their apps works in Chrome.   Who does this ?

And why aren't the devs fixing things that this contributor has provided, including instructions on how ?   He is making software to make DCS a great experience in VR, I think you'd find some value in that.

The irony is this patch was delayed - maybe it should have been delayed longer - I can't use my VR and I can't use my FC3 modules I paid for.   And my F-15E isn't getting updated anymore.  What is going on ED ?

[Morphine]

2 hours ago, mbucchia said:

@BIGNEWY this is the same bug I reported to you guys over a year ago, with incorrect FOV submissions in xrEndFrame().

There is a workaround to YOUR bug in Quad-Views-Foveated, but it looks for the application name being "DCS World", but it looks like you changed that string to "DCS", so the workaround doesn't kick in.

Please fix the real issue anyway... your code is violating the OpenXR standard.

Reported originally March 2023 with a very clear explanation (point 2):

 

 

@mbucchia BTW, this same bug currently makes the Varjo native DFR unusable with the same problem which is why everyone is still using your Varjo Foveated software. Hopefully DCS fixes this and that will enable us to use the Varjo Native DFR.

Edited July 11, 2024 by Morphine

[mbucchia]

2 minutes ago, Morphine said:

@mbucchia BTW, this same bug currently makes the Varjo native DFR unusable with the same problem which is why everyone is still using your Varjo Foveated software. Hopefully DCS fixes this and that will enable us to use the Varjo Native DFR.

 

I assume you mean "Varjo native DFR" as in the new registry key they've released in 4.3. That would be correct, the Varjo runtime doesn't do the incorrect FOV submission "recovery" that I added in Varjo-Foveated and Quad-Views-Foveated.

That's because this workaround is super sketchy (we guess which dynamic projection the game should have submitted), it uses a FOV cache to try to re-correletate the FOV information with the previous frame(s). This is why I only activate this if I am sure the app is DCS, since this is fragile code that could break other apps intentionally doing FOV manipulation.

 

Note that this isn't an uncommon bug that ED have in their code: there's quite a few apps doing incorrect pose/FOV submission, but most of the time it only affects ever-so-slightly the reprojection (or adds a bit of instability). Here with the very low latency eye tracking, the effect is absolutely dramatic.

[Morphine]

@mbucchia

Yes, I mean the new Registry key in 4.3.

So this is something DCS needs to fix and not a Varjo issue or both?

[mbucchia]

8 minutes ago, Morphine said:

@mbucchia

Yes, I mean the new Registry key in 4.3.

So this is something DCS needs to fix and not a Varjo issue or both?

Not a Varjo issue. 

DCS submits information incorrectly to OpenXR. My original post from March 2023 had a screenshot of the OpenXR calls that made the problem very obvious, but somehow the forum doesn't have the image anymore and it shows as a broken URL...

Edit: ugh the broken link is my fault, my image file was hosted on Discord, and with the deletion of my server, it's gone.

Edited July 11, 2024 by mbucchia

[BIGNEWY]

folks the issue is reported to the team.

I get you dont like the way we do things or you think it could be done better, I have given all the feedback to the teams. 

Keep it constructive, and on topic thank you. 

[Mainstay]

Pimax Crystal user here... no DCS for me till its fixed.

@Eagle Dynamics Lets go ED please fix this and take Mbucchia's notes into account please.

[tikijoetots37]

10 minutes ago, mbucchia said:

Not a Varjo issue. 

DCS submits information incorrectly to OpenXR. My original post from March 2023 had a screenshot of the OpenXR calls that made the problem very obvious, but somehow the forum doesn't have the image anymore and it shows as a broken URL...

Edit: ugh the broken link is my fault, my image file was hosted on Discord, and with the deletion of my server, it's gone.

 

Hi mbucchia!, thanks for explaining everything. I went through and read all your posts regarding this. Do you think this would be something that would be complicated/time consuming for ED to fix the right way like you explained in earlier posts.. ie for DCS to OpenXr to the right standards? I guess I ask to see what kind of time frame we might be looking at. I'm going to go ahead and assume that if its an easy fix we'll get it sooner rather than later. My fear is that the difficulty or time for them to do it the right way might not be worth it to them for what I assume is a small portion of their customer base. I understand you might not have the answer for the above, just thought that your knowledge of it might have some insight on the complexity of the fix. Best regards and thanks!

[Glide]

7 minutes ago, Mainstay said:

Pimax Crystal user here... no DCS for me till its fixed.

I'm on the Crystal as well, and I say give it a try without QV.  With my 3080Ti and the headset at 66% resolution I'm getting 60fps with everything turned up max.  Perfection.

No QV, no Turbo Mode, no Low Latency, 120hz refresh rate, no smoothing or headset foveated rendering.  I even turn off eye tracking and position reminder.

[Hawkeye_UK]

11 minutes ago, Glide said:

I'm on the Crystal as well, and I say give it a try without QV.  With my 3080Ti and the headset at 66% resolution I'm getting 60fps with everything turned up max.  Perfection.

No QV, no Turbo Mode, no Low Latency, 120hz refresh rate, no smoothing or headset foveated rendering.  I even turn off eye tracking and position reminder.

66% resolution - yikes, i might as well just connect up my Rift S lol......thanks for the tip and yea i've tried but for my eyes it felt like going back a generation in VR dropping the res down that much.

The reality was this was warned about - and well ED didn't take that into account.  There is no stable, and no warning.  There should have been on the patch notes this is not compatible with high end headsets using Quad Views (an accepted norm these days by most in the community).

A situation that should not have occured.  I personally would not have hit update today if i had have known in advance.  

[javelina1]

I tried launching via mt, and have been at a white screen for about 10 minutes.  I hear the F4 theme music jamming away.  I have no mods installed, just vanilla.  trying to get an image on my Aero, but nada....

earlier, I had checked that the toolkit did not have DCS checked.  but I just went in again and looked, and there's the DCS and DCS World.  so I made sure both were not checked (to be enabled).  I'm now able to launch into VR.  But unfortunately, cannot do any further testing until tomorrow...   

Edited July 11, 2024 by javelina1
added info

[Ready]

I was looking forward to the announced performance increases because of AI changes and was pleasantly surprised by a fancy looking loader. But unfortunatly I also ran into this issue (Aero here). 

Turning off QVFR and reverting to other eye tracking software for me means impact on my recordings.

This would definitely have been caught during the testing, if the release was tested with this configuration, which I think is currently pretty mainstream. I previously understood there were enough VR testers so maybe it was caught and impact was accepted due to time pressure. If things like this get spotted during the testing and get pushed to production, could they also be pointed out in the release notes? That would have saved me a load of time, headscratching, restarting, reinstalling, looking at discord, recording a video to post the bug and finally coming to this thread.

It anyhow is dissapointing to see mbucchia's earlier notifications and instructions falling on deaf ears. 

Any ETA on the fix?

Edited July 11, 2024 by Ready

[Glide]

30 minutes ago, Hawkeye_UK said:

yikes

I can go higher if I use DLSS.  That's with TAA.  I just did the Beruit test and the Lebanon mach loop in the F16, and wow, I'm amazed.

[mbucchia]

57 minutes ago, tikijoetots37 said:

Do you think this would be something that would be complicated/time consuming for ED to fix the right way like you explained in earlier posts.. ie for DCS to OpenXr to the right standards?

I can't speak for ED, my guess is just like any software team, the more consuming part isn't the code change itself but the review, validation and rollout process that comes with it.

From the code change itself, I don't know what their code looks like, but I can't imagine that them storing the proper FOV values (they retrieved from xrLocateViews() to compute the projection matrix) into a variable that follows the frame's lifetime until submission (via xrEndFrame()) would be complex. The code in QVFR and VFR does that, and it's something like 20-30 lines of code that were written in well under an hour.

[mmike87]

Good grief ...  I haven't downloaded the new version yet, comma, and at this point, I may just create a whole nother install. I'm on an Aero but have always used Varjo Foveated, So hopefully it will be fine.

Love this game ... But is always three steps forward, two or more steps back with most releases. 

[Archer77]

@mbucchia I'm willing to pay 1000$ for to get a certificate for Quad Views. Let me know.

Edited July 12, 2024 by Archer77

[diego999]

Going back to the previous version. DCS is unusable for me in this state.

[mbucchia]

3 minutes ago, Archer77 said:

@mbucchia I will willing to pay 1000$ for to get a certificate for Quad Views. Let me know.

Appreciate that but obtaining a signing certificate would take like 2 weeks (the whole idea of it is that they verify you are the person who you say you are, so it's a slow process, sometimes might require to visite a notary). Then on top of it, there are several hours of work to setup the new signing process on the cloud.

My hope is something that simple and easy will be fixed by then.

Meanwhile there are multiple solutions:

- Use DCS ST (single-threaded), make sure to run with `--force_OpenXR`)

- If you are on Varjo, you can revert to using Varjo-Foveated instead of Quad-Views-Foveated. 

[SneakyBastd]

Also having quad views problems here with Crystal/4090

It's really bad on some modules like Apache and Viper.

Almost normal with hornet and some other aircraft. Weird.

ED, please fix this ASAP. And for gods sake test your bloody game in VR !!!!!

And thankyou AGAIN mbucchia, tried ST. It does indeed fix the problem. But the game still looks quite average as you can't enable DLAA etc.

Edited July 12, 2024 by SneakyBastd

[nikoel]

Woah. We have some high rollers in ere'

I am also going to constructively say that I really hope that the feedback that you're passing on @BIGNEWY is taken onboard rather than "yes dear; thank you dear"

Quadviews is as much of an advancement for VR as Multithreading - this is a big deal to many of us

I have fixed this with the patch made by @sandboxcode - so from the bottom of my heart thank you mate. It would be good to have something a little more official from DCS as this fix screws up all other games that rely on anticheat and QuadViews

Edited July 12, 2024 by nikoel

[mbucchia]

15 minutes ago, nikoel said:

It would be good to have something a little more official from DCS as this fix screws up all other games that rely on anticheat and QuadViews

To be very clear, the unsigned DLL will screw up ALL OpenXR games that have anti-cheat, regardless of whether these games use or not use quad views.

It will make them fail in a way that will also pop up my name on the screen "failed to load XR_MBUCCHIA_quad_views...." aka pointing fingers at me

This is why I'm not distributing the unsigned DLL.

Some examples of games using OpenXR and Anti-cheat that will break with an unsigned DLL system-wide: Pavlov, WRC, Contractors Showdown, Roblox...

 

EDIT: FYI this is the bill for getting a digital cert in 2024. Per year. We can thank the anti-cheat companies for that. I wish Epic and friends would pick up that bill...

null

Edited July 12, 2024 by mbucchia

[Chief_Biv]

Crystal/Quadviews user here who was lucky enough to not update DCS yet. I would like to know from ED if there is a short or long wait to fix this too? I suppose I should refrain from updating for now, but was looking forward to using the new patch's content.