2023.3.0 - Relative Translation, Tobii 5 and DCS; incorrect axis?

[Keebler750]

I'm LOVING the stability of tracking now that I'm using this setup, but there doesn't appear to be any documentation for the rather cryptic RT parameters and I'm having a tough time troubleshooting.

I have to ask - what is 'relative translation' supposed to be relative to?

When using the native Tobii apps, relative translation is relative to the current view angle so that when you look to the side and move forward, your ingame view moves toward the view vector.

When using OpenTrack, the translations appear relative to the nose of the aircraft no matter what I try to do. When I look to the side and move my head forward as above, you can see the camera shifting toward the front of the plane, not toward the view vector.

Is this a setup problem on my end, or is this a known bug?

INPUT: Tobii Eye Tracker
OUTPUT: freetrack 2.0 enhanced; Track IR disabled.
FILTER: Hamilton
(Tobii Game Hub exited in System Tray)

[Keebler750]

After reading and researching, it is clear this is coded to flip some outputs depending on view angle. That is NOT the same as relative translation, and should not be labeled as such. Perhaps this has happened because the feature was requested based on 'looking behind an aircraft' instead of a more technical design related to camera position/rotation/zoom.

Relative translation is the binding of the head movements in the same way a human head works, namely: when you turn your head, and translate your head to your HEAD'S left, the view shifts relative to where it is pointed, not relative to the chair (or airplane) you are sitting in.

What this means is that the view vector becomes the reference, instead of the object we are piloting, and that is why the yaw/pitch/roll is processed first, and then translated in the RELATIVE x/y/z.

If you look at Tobii 5 output, their relative translation is coded as the human head would work, and it has been decoupled from the vehicle axis. This feels MUCH more natural and intuitive, since this is how we use our head and eyes in real life.

Something is happening in the OpenTrack code which continues to couple to the vehicle axis. The camera position output needs to couple to the view vector for actual 'relative translation.'

I suspect this Discussion should be made an Issue.