Head tracking - yaw effect on z-axis issue

[imacken]

Has anyone found a solution to the issue of the effect that yawing has on the z-axis when using a head clip attached on the side of the head?

By that I mean that when you sit and just turn your head left and right, as well as moving left and right, the image moves slightly in and out as the sensors are actually getting closer and further away in the turning process.

(This doesn't happen when using the TrackIR hat clip - for obvious reasons.)

[imacken]

Anyone?

[Len62]

You can try and increase the dead-zone on the z-axis.

[finch]

I think the problem is trackir not properly configured and not DCS

[imacken]

I'm not saying there is anything wrong with TrackIR or DCS. The fact is that when you turn your head with a tracker on the side of your face, then the LEDs do move on the z axis. That's because they are on the side, and that is why it doesn't happen when you use a hat clip as the LEDs are centred.

I'm just wondering if anyone has found a technique/setting that can overcome this, or at least minimise the effect. Increasing the dead zone on the z is not really the answer as that has other implications.

[Mecrutio]

TrackNoIR has a calibration function that addressed this issue. Maybe your software has the same?

[dwpenney]

Just a guess but I would assume that you are moving your neck when attempting to 'roll' on the z-axis. Basically there are two ways to roll your head: 1) at the base of the skull; 2) farther down your neck where it meets your shoulders; With a Track Clip Pro, you are correct that it is on the side and will induce some level of yaw. You can mitigate this by rotating from the base of your skull as the TrackIR software takes into account this distance from the rotation point. You can try what I am talking about in the TrackIR software as it displays the axis and the disembodied heads. With the track hat the same issue should exist .. just in a different axis. The farther the sensor target is from the centre of rotation the more of an affect it will have on other axis.

[imacken]

Thanks for that. No, it’s not really the way the head is moved that I am talking about. I am talking about the effect that yaw with side mounted headsets has on the Z axis.

In TrackIR, it’s easy to see. Just look at the top view in the software and yaw left to right. You can then see the sensors moving above the line when looking left, and below the line when looking right. The ‘real’ head doesn’t do that in the same view.

As Mecrutio said above, TrackNOIr has a calibration tool which circumvents this issue. Opentrack has this as well.

I cannot see how to do this in TrackIR.

[imacken]

Just in case I'm not explaining myself very well here, I have attached a graphic showing what I mean.

In it, you can see the virtual head in various views for a 90 degreee left and right yaw for both the head clip and the hat clip.

Hopefully, you can see that the head clip is perfect, with the head positioning exactly how you would expect. If you then look at the head clip shots, you will see what I mean. The issue is most obviously seen in the 'top' view where the head is forward and back of where it should be, but it is also visible in the 'left' view as well. This is not unexpected as the view is pivoting around a point that is off to the side when wearing a head clip.

I feel that I must be missing something in TrackIR settings, as this issue is easily resovled in Opentrack and FaceTrackNOIR by a calibration option.

Any further thoughts, guys? If not, I'll contact Natural Point support and ask them for their thoughts.

[Chic]

You might be able to mitigate the Z axis effect by repositioning the TiR receiver.

[imacken]

You might be able to mitigate the Z axis effect by repositioning the TiR receiver.

Thanks but it’s not that. It’s a simple physical fact that with the side head mounting, a turn to the left or right will obviously move the z axis sensor in or out, which in turn distorts the actual movement. This does not happen with a head mounted centre positioned sensor.

[Chic]

Thanks but it’s not that. It’s a simple physical fact that with the side head mounting, a turn to the left or right will obviously move the z axis sensor in or out, which in turn distorts the actual movement. This does not happen with a head mounted centre positioned sensor.

 

 

 

Understood, but, I believe that by repositioning the sensor laterally you can influence the relative Z axis excursion. You can only mitigate Z axis activity, not remove it. Z axis activity is generated with a head mounted centre positioned sensor as well. It's just to minor to measure.

I'm sitting 28" from a 65" TV monitor with camera mounted on top so the TiR setup angles are pretty severe. Despite this, I have managed to negate any relevant distortion by manipulating position and angle of sensor mount.

Please post if you get a useful response from Natural Point support.

Edited October 27, 2017 by Chic

[imacken]

Z axis activity is generated with a head mounted centre positioned sensor as well. It's just to minor to measure.

Exactly, that's the point, it's too small to measure with a centre mounted device. Look at the attachment I posted above.

What I don't get, as I have said before, is that (the free) Opentrack and TrackNOIR have a calibration tool that elimintes this issue, but I'm pretty sure that (the certainly not free!) TrackIR doesn't, unless I'm missing something.

[Chic]

but I'm pretty sure that (the certainly not free!) TrackIR doesn't, unless I'm missing something.

 

I'm sure you're aware that you can decrease Z axis sensitivity in settings, but you that would also make it harder to zoom when you want to.

Other than that, I'm unaware of any other means of calibrating this issue.

Edited October 27, 2017 by Chic

[imacken]

I'm sure you're aware that you can decrease Z axis sensitivity in settings, but you that would also make it harder to zoom when you want to.

Other than that, I'm unaware of any other means of calibrating this issue.

I'm going to contact Natural Point on this, as it seems crazy that free software can handle it, yet TrackIR can't.