OpenTrack drifting and recentering issue

[lovelyhead]

Can you confirm you updated your Rift firmware to 0.18?

Yes I updated it on the same day the updated SDK came out.

Opentrack crashing when you stop the tracking isn't a big deal to me. I'm just glad it works!

[Tankshell]

Just tried this latest version/build, at first I thought it was working, but after a few races on Dirt 3 the yaw drift is still quit bad. By the end of the race my default view was pointing waaaaay to the left. So I don't think this is fixed still.

Is this the DLL that drifts to centre like the one above??

I can get the centre drifting one to work but when i copy the dll stuff over the opentrack folder with that latest dropbox upload, the rift options stops being an option in opentrack

I had the recenter spring flag turned on when I did the build so yes, I think it does do the recenter.

That's odd about this one not working for you. I have it running fine. When you look in the opentrack folder you see all the rift files with a date of 10/14? You have opentrack v2.0a2?

EDIT - upon closer testing, it appears to be the "drift to center" feature of this new build that is screwing up. Can you do a build without "drift to center"? If the yaw drift is indeed fixed now, we don't need it anyway...

To test this, if you run opentrack, and hit start... then rotate the rift around a bit as if you were playing a game, then place your rift flat on the desk and don't touch it keep an eye on the yaw value, you will notice that it incorrectly increases/decreases the yaw value constantly after a short while! It seems like the auto drift to center is not taking into account the new magnetic drift correction, and in essence is now fighting it...

Either that or something terrible has gone wrong with my Rift...

[sthalik]

Hey guys,

wrt MSVCRT dependency hell, check out the free 'dependency walker' program, it shows conflicting MSVCRTs among other DLL issues.

multiple CRTs are troublesome because they have different struct sizes and as such, passing them across object boundaries won't work.

Some people use multiple CRTs to great success, but only if platform-specific stuff like stdlib, stl are avoided compeltely. Even cstdio has to be avoided...

What's status of Rift support with 0.2.5? Can I update the build machine with SDK 0.2.5 and will it work without any issues?

Keep in mind, got no actual device for tests. Is there an emulator or something for developers?

cheers,
-sh

[sthalik]

Can someone clone this:

https://github.com/opentrack/opentrack/ ... bovr-0.2.5

and see if it attaches to sensor and works somewhat? Didn't see mm0zct today online, so did the work myself. Hope he doesn't mind me stepping on his turf...

[mm0zct]

Good news, sorry for my absence but PhD work took over my dev time recently, sthalik and I have updated Opentrack to use the 0.2.5 version of the sdk and drift seems to be gone at last. The app doesn't crash on exit or anything either any more.

The official alpha can be picked up at http://ananke.laggy.pk/opentrack/ (2.0a6 is the current one at time of writing, but I'd probably recommend grabbing the newest).

The curves etc will be set up wrong just now (defaults to monitor based exaggerated headtracking), so you should go through the curves and make them all straight lines, and set the filter option to "none".

Please remember to update your Rift firmware and do a magnetic calibration profile using the SDK utility.

[vulkan]

Thank you mm0zct!

May I humbly request a minor feature, is it possible to monitor a game controller (i.e. joystick button) for re-centering?

[sthalik]

No joystick/mouse keybindings available.

This is feature bloat, given how elephantine windows APIs are.

Also, it requires more headers than it usually does, to support it on Linux as well. Code bloat, to have two widgets, one on Windows, one on Linux.

[mm0zct]

Since I'm currently looking at refactoring the hotkey system, I might look into it, but I can't make any promises. If the drifting is solved why would you need a recentering hotkey anyway? (other than "just in case")

[sthalik]

While it's not a user-visible change, I'd make a statically-typed settings template.

But maybe it's too late in the release process...?

How's testing Linux libovr?

Is there much need for Apple Macintosh support?

[Mattijs]

It works for me now with the release version!
Thanks!

Pushing the centering button does make it crash though and when that happened the rift option disappeared again and didn't come back after a restart.
I had to delete all the opentrack stuff en redo the copying of the DLL over the new folder.
When it crashes now, which does happen when you push recenter and on some occasions but i can't figure out which for the moment, it keeps the RIFT option as a tracker after just restarting the app.

With the magnetic calibration it seems i always have to look a little too much upwards to look outside the window of the car in DIRT.
I used to have this but just recentered around what felt good.

It might be that the driver isn't as tall as i am though.

There are dails for controlling the Z, X ,Y through opentrack's trackIR port...but they don't seem to work. I don't have a hydra so i can't cheat by calibrating the hydra when i'm lower and then drive with my head higher so i'm closer to the roof.

Is there an easy fix for that?

[sthalik]

There isn't a single reason why it should crash. Probably a problem with drivers.

If you want to help, get us a backtrace.

[mm0zct]

On the driver height issue in Dirt, the same utility which you use to change the in-car FOV can be used to nudge the driver head position in each direction, I have mine set to move the camera a bit higher. If you haven't changed the FOV then I'd recommend this as a very important step (see the Vireio thread on Dirt2/3).

In OpenTrack ahve you set all of the curve controls properly? they should all be straight lines from bottom left to top right, and the "alternative" checkbox should be unckecked, which it's not by default for pitch (looking up/down). Once the curves are set up and I've run the FOV changer utility I have no problems using the Rift in Dirt2.

[vulkan]

Since I'm currently looking at refactoring the hotkey system, I might look into it, but I can't make any promises. If the drifting is solved why would you need a recentering hotkey anyway? (other than "just in case")

After prolonged use I find it occasional gets out of center, even the new version. I'm playing a WW2 online flight sim (Aces High) and the head movements are fairly insane in combat .

Of interest I have some SteelSeries headphones that cause tracking drift when I wear them (ironically I won them from MTBS3D!).

[mm0zct]

Interesting, I'll whack a yaw drift-to-centre back in with a gui option then, probably try and give you access to the spring and drift constants too so you can make it stronger/weaker as desired.

[sthalik]

mm0zct, any update, or are you still busy?

[vulkan]

FWIW Rift+Hydra+Tridef+OpenTrack in action.

http://www.youtube.com/watch?v=0q51xXnLgj4

I have found that my headphones (Steelseries) seem to be causing drift, with the latest version/firmware of Opentrack and NO headphones I'm not seeing the drift I was.

[mm0zct]

I would hazard a guess that the magnets in your headphones are confusing the Rift, did you calibrate wearing them or without them? If you calibrated without I suggest you try creating a calibration profile while wearing them and see if you still suffer drift while wearing them.

Sthalik - I'm about to test in Linux for Rift and hopefully Hydra, I'll message you when I have anything to report.

[vulkan]

I cannot calibrate, it will not complete for me (with or without headphones).

I purchased some cheap (less powerful magnets) headphones and don't have the same issues. Ideally some in ear units would do the job BUT I cannot find any with a built in mic.