Head position randomly self-resets to an arbitrary incorrect position

[Aquorys]

Without any apparent reason, the head position suddenly jumps to a different and nonsensical location in the cockpit. This seems to happen with all aircraft (so far, I have tested with the F-16 and the Su-33), and the location that it jumps to is the same every time, but is a different for each aircraft (e.g., into a forward-left position of the cockpit in the F-16, straight forward way in front of the HUD in the Su-33).

The position jump is not relative to the current head position, it is absolute, meaning that even if I intentionally reset my headset position so that I end up in a different position in the cockpit while flying, I still end up in the same location after the sudden jump. I tested this by resetting my position so that I would fly the aircraft from behind the seat (as if I were in a backseat).

On average, it happens every 10 minutes - sometimes it would happen twice per minute, sometimes it would work for 20 minutes before failing again.

The headset is an Oculus Rift S. To rule out the possibilty that this is yet another problem of the extremely buggy Oculus drivers, I flew a turboprop around for over an hour in the same region in another flight simulator, but encountered no VR problems at all, so I am convinced that this bug is in the DCS VR code.

I first encountered this problem a couple days ago, so it may be caused by one of the last updates. I tried "recenter VR headset", resetting the camera position, saving the cockpit angles, and a few other view-related things, but all to no avail.

I have attached a video capture of what the problem looks like.

[Aquorys]

I just found this thread from 2019 that seems to describe the same problem, or at least a very similar one. I have to say that I find it very frustrating that obviously nothing was done over the course of the last 3 years to fix such an obvious programming error.

[dutchili]

Same here with Oculus Quest 2.
For me, this problem started at the beginning of the week after:

- updating windows 11 using H2 release

- updating Gerforce 3070 driver

 

The problem seems to fade after a few minutes and primarily occurs when looking to the far left, right or down.

In general (VR) performance is down after yesterdays small update.

 

[Aquorys]

tl;dr: The problem is caused by an uncommanded realignment of coordinate systems by DCS, explanation below.

I could reproduce the same problem that I had with intermittent tracking problems by taking the headset off, waiting for it to reset itself, and then putting it on again. In both cases, the position tracking is lost intermittently and later reestablished, and the effect on the head position in the cockpit is the same.

There are two sets of coordinates, one in the game world and one that is reported by the VR headset.

Let's call the head position coordinates in the game world [Hx, Hy, Hz] and the VR coordinates [VRx, VRy, VRz], and let's assume that the normal head position in the cockpit is at coordinates [Hx=0, Hy=0, Hz=0]. For a normal cockpit view in VR, whatever coordinates the VR headset is reporting when the player's head is in its normal position must be aligned with the game coordinates. This is also essentially what the "View - Recenter VR headset" function does.

In my case, the VR coordinates for my normal head position in the cockpit are roughly [VRx=-0.17, VRy=-0.315, VRz=-0.589]. These coordinates are aligned to [Hx=0, Hy=0, Hz=0] when everything is working properly (in other words, [VRx=-0.17, VRy=-0.315, VRz=-0.589] == [Hx=0, Hy=0, Hz=0]).

Here is a picture of the initial position in the cockpit, with VR coordinates [VRx=-0.17, VRy=-0.315, VRz=-0.589]:

Then, whenever the position tracking is lost, the headset stops reporting any (valid) coordinates. When the position tracking is later reestablished by the headset, the headset starts reporting valid coordinates again:

As it can be seen in the picture, the coordinates are again very close to the ones that were reported before the position tracking was lost intermittently. Therefore, one would assume that the head position in DCS should be roughly where it was before the intermittent tracking loss.

But this is not the case, because DCS has realigned the two coordinate systems so that [Hx=0, Hy=0, Hz=0] == [VRx=0, VRy=0, VRz=0], which makes no sense, because there is no reason to assume that the headset coordinate system has changed.

This is easily confirmed by moving away from the normal player position to reposition the VR headset as close as possible to [VRx=0, VRy=0, VRz=0], which looks pretty much like a normal cockpit perspective again:

The solution to this problem is to change the DCS VR code so that no uncommanded realignment of the two coordinate systems is performed.

PS: As for why the headset would even lose track when it is not being taken off, just because its cameras can't find enough "trackable objects" and "corners", I can't tell - that thing has accelerometers, gyroscopes, a magnetometer, and several cameras. The cameras should be used for finding the initial position and for the correction of gyroscope drift, but the headset should otherwise mainly rely on inertial sensors and should easily be able to go for quite a long time with minimal drift without using the cameras. I have used gyroscopes under way worse conditions (high acceleration, high vibration environments) and I would expect them to fail maybe every 4 years, not 4 times in 30 minutes. How the headset developers managed to make these sensors less reliable by a factor of roughly 5 orders of magnitude is beyond my imagination.

[draconus]

The only problems with positioning in Rift S I had were always when the room was not lit enough. When the monitor is on it provides just enough light to track the front angles view but tracking can be lost whenever you turn your head more to the sides. When that happened I was invited with "playspace area not set" info and was forced to set it up again. I wish it was just sudden reposition that I could reset in a moment. So my advise is to make sure the room is lit enough and try to setup the playspace area when sitting exactly the same when you play DCS.

[Aquorys]

Thanks for the information. I think light should not be the problem, I can try it again during daytime with sunlight, but that room isn't too large and the light source is the LED equivalent of a 100W light bulb. With regards to the position data, it seems that the zero position for the Rift S is not calculated from the seat position, but from the geometric center of the "Guardian" area that you draw into the room. Kinda hard to get that perfectly centered around the seat.

[draconus]

I always cancel the guardian setting fyi.

[DST]

I´m glad somebody finally adressed this, because I have this problem too for quite some weeks now, and it appears to be very annoying. Every 10 minutes or so the headset seems to loose tracking for a split second and then view re-locks too far down or too far behind, so I have to constantly push Num5 for recenter view. First I thought it is a hardware problem with headset (Rift S user here) but it definitely occurs only in DCS, and only since a few weeks after last or second last update before 2.8.

It would be very nice if the devs could have a look into this, as it really makes it hard to enjoy DCS for me at the moment.

[DST]

On 10/11/2022 at 10:27 PM, Aquorys said:

tl;dr: The problem is caused by an uncommanded realignment of coordinate systems by DCS, explanation below.

I could reproduce the same problem that I had with intermittent tracking problems by taking the headset off, waiting for it to reset itself, and then putting it on again. In both cases, the position tracking is lost intermittently and later reestablished, and the effect on the head position in the cockpit is the same.

There are two sets of coordinates, one in the game world and one that is reported by the VR headset.

Let's call the head position coordinates in the game world [Hx, Hy, Hz] and the VR coordinates [VRx, VRy, VRz], and let's assume that the normal head position in the cockpit is at coordinates [Hx=0, Hy=0, Hz=0]. For a normal cockpit view in VR, whatever coordinates the VR headset is reporting when the player's head is in its normal position must be aligned with the game coordinates. This is also essentially what the "View - Recenter VR headset" function does.

In my case, the VR coordinates for my normal head position in the cockpit are roughly [VRx=-0.17, VRy=-0.315, VRz=-0.589]. These coordinates are aligned to [Hx=0, Hy=0, Hz=0] when everything is working properly (in other words, [VRx=-0.17, VRy=-0.315, VRz=-0.589] == [Hx=0, Hy=0, Hz=0]).

Here is a picture of the initial position in the cockpit, with VR coordinates [VRx=-0.17, VRy=-0.315, VRz=-0.589]:

Then, whenever the position tracking is lost, the headset stops reporting any (valid) coordinates. When the position tracking is later reestablished by the headset, the headset starts reporting valid coordinates again:

As it can be seen in the picture, the coordinates are again very close to the ones that were reported before the position tracking was lost intermittently. Therefore, one would assume that the head position in DCS should be roughly where it was before the intermittent tracking loss.

But this is not the case, because DCS has realigned the two coordinate systems so that [Hx=0, Hy=0, Hz=0] == [VRx=0, VRy=0, VRz=0], which makes no sense, because there is no reason to assume that the headset coordinate system has changed.

This is easily confirmed by moving away from the normal player position to reposition the VR headset as close as possible to [VRx=0, VRy=0, VRz=0], which looks pretty much like a normal cockpit perspective again:

The solution to this problem is to change the DCS VR code so that no uncommanded realignment of the two coordinate systems is performed.

PS: As for why the headset would even lose track when it is not being taken off, just because its cameras can't find enough "trackable objects" and "corners", I can't tell - that thing has accelerometers, gyroscopes, a magnetometer, and several cameras. The cameras should be used for finding the initial position and for the correction of gyroscope drift, but the headset should otherwise mainly rely on inertial sensors and should easily be able to go for quite a long time with minimal drift without using the cameras. I have used gyroscopes under way worse conditions (high acceleration, high vibration environments) and I would expect them to fail maybe every 4 years, not 4 times in 30 minutes. How the headset developers managed to make these sensors less reliable by a factor of roughly 5 orders of magnitude is beyond my imagination.

So would you say the problem is the headset? Because I don´t have this issue with other flight simulator or VR games. It also seems that this issue correspondends with moments of heavy fps loss, like carrier approach or moving around on carrier.

[Rufuz64]

Perhaps it is not only the problem of having too little light, but also too little contrast. A white wall would be bad, something like this would probably be optimal for good tracking:

null

Edited November 2, 2022 by Rufuz64

[DST]

24 minutes ago, Rufuz64 said:

Perhaps it is not only the problem of having too little light, but also too little contrast. A white wall would be bad, something like this would probably be optimal for good tracking:

null

 

Like I said above, the tracking of my headset runs perfectly fine with every other VR application, I'm pretty sure this is a DCS related problem with something they added in one of the last few updates. I would like to know if this problem is encountered by reverb users too.

[parasxos]

Have the exact same problem in DCS only. Rift S will loose tracking for a second and then my cockpit view will be outside of the plane. Never happens in all other vr sims I'm using. Didn't have this issue before...

[Bulldog_1]

Same problem. HP Reverb Pro. Open XR or Steam VR.

[Sickdog]

Myself and my friend are having this issue as well, I want to say mostly after 2.8 but really can't say for sure as neither of have played much for the past month. Maybe a Windows update? He's running an original Reverb and I'm using a G2, both with Windows Mixed Reality of course, but also we're using OpenXR. Can't say we see issue in other applications or not as we pretty much use VR with DCS exclusively, but I can say this wasn't a problem I noticed until recently. 

[RuskyV]

I have the same issues with the rift-s, most of the time the view will jump outside the cockpit and be facing backwards, might happen once or twice during a multiplayer session of 1hr or longer.

[Chipensaw]

Good morning,

Any update with this issue?  I have the same issue with Quest 2.  Issue started with update to 2.8.  Specs and setup below.

5800X3D
3070ti
32G RAM
Dedicated M.2 SSD for DCS

No mods, clean install.  No 3rd party VR programs in use, FXO & Meta deleted.  PD1.3, textures high, terrain textures low, shadows off, MSAA 2X, AF 16x, clouds high, other non related settings available upon request.  @BIGNEWY 9L asked me to reach out to you regarding VR performance with my setup.

[BIGNEWY]

4 minutes ago, Chipensaw said:

Good morning,

Any update with this issue?  I have the same issue with Quest 2.  Issue started with update to 2.8.  Specs and setup below.

5800X3D
3070ti
32G RAM
Dedicated M.2 SSD for DCS

No mods, clean install.  No 3rd party VR programs in use, FXO & Meta deleted.  PD1.3, textures high, terrain textures low, shadows off, MSAA 2X, AF 16x, clouds high, other non related settings available upon request.  @BIGNEWY 9L asked me to reach out to you regarding VR performance with my setup.

Hi we will take a look, and chat in discord, thanks 

[Tomcat388th]

you can add me to this list as well G2

[BIGNEWY]

Please check
Untick use snap views in DCS options 
Make sure your saved games dcs config view folder is clear.

Make sure if you have ever used the cross eye tool you dont have any old VRheadset.txt files in your saved games dcs folder. 

If you still have problems with head position after it maybe headset tracking issue

[Baldrick33]

I have previously had some tracking issues (Reverb G1) that were resolved with my hardware. USB connection in one case, replacement G1 cable the other.

The reason I think it can seem like DCS because it works fine in other sims is the issue is more apparent when the PC is on its limits - the PC reaches a point where it momentarily loses connection to the headset and the tracking resets. In my case I had the same issue with iRacing.

One other thing to check is whether you have game mode on in Windows. Whilst this prioritises resources to the game in some cases it can take too much away from the VR software, again causing a temporary loss of connection in heavy usage scenarios.

[nickos86]

I played ALYX with a G2 for HOURS without any alignment problems (both in a dark and bright room, with a lot of movement). So I do think it something that should be mitigated via DCS or perhaps WMR?

When turning the G2 ON, I try not to touch the hardware and let it calibrate with as little interference as possible. It seems to help with the issue a bit. 

But eventually, I think it's more of a software issue that should be resolved. 

[Hoirtel]

This is happening to me in quest 2. Started with 2.8 sometimes not too bad, sometimes terrible. I will try and get a track. Hope there is an easy solution as this is really difficult to use. 

[Baldrick33]

17 minutes ago, Hoirtel said:

This is happening to me in quest 2. Started with 2.8 sometimes not too bad, sometimes terrible. I will try and get a track. Hope there is an easy solution as this is really difficult to use. 

I still think it is load on your PC causing the tracking to not be able to keep up for a split second. 2.8 has increased the load for VR compared with what you were running and maybe tipping it over the edge. You could try reducing some settings and see if it stops the tracking issues?

If 2.8 had introduced a tracking reset bug I would expect it would be happening for everyone

[Hoirtel]

1 hour ago, Baldrick33 said:

I still think it is load on your PC causing the tracking to not be able to keep up for a split second. 2.8 has increased the load for VR compared with what you were running and maybe tipping it over the edge. You could try reducing some settings and see if it stops the tracking issues?

If 2.8 had introduced a tracking reset bug I would expect it would be happening for everyone

I think that still classes as a bug! I'm not experiencing poor performance in these situations so if the is the case it's really split second. I'm not denying that could be what's happening but reducing "some settings" isn't an answer that I can work with really. It is also happening for others but cropping up on other threads. I found a link to this one. 

[Baldrick33]

45 minutes ago, Hoirtel said:

I think that still classes as a bug! I'm not experiencing poor performance in these situations so if the is the case it's really split second. I'm not denying that could be what's happening but reducing "some settings" isn't an answer that I can work with really. It is also happening for others but cropping up on other threads. I found a link to this one. 

By reducing settings I meant simply to see if that fixes the tracking issues. If it does then we need to dig deeper into priorities of resources.