Page 1 of 1

Positioning and reference grid

Posted: Mon May 18, 2009 5:30 pm
by simon_o
I don't understand how the coordinates work.

I put the vector clip on the ground at say, 12 inches from the sensor. I get a certain z value (using the OptiTrack SDK). Then, I move the clip left and right, although I make sure to stay at the same distance along the z-axis, and my clip is always facing forward, i.e. along the z-axis (NOT toward the sensor). This makes my z value change by up to 40 mm. Why does it do that? Lens geometry that is unaccounted for?

Also, I noticed that at a same position, if I just rotate the clip around the y axis (yaw rotation) where the rotation point is the point halfway between the 2 front reflectors of the clip, my x and z values change radically. Does this mean that the x, y, z reference grid follow the orientation of the clip? Or is the x, y, z grid relative to the sensor?

Re: Positioning and reference grid

Posted: Tue May 19, 2009 1:31 pm
by Birch
If you are not using a TrackIR3/TrackIR4 you may want to update the lens parameters to get more accurate readings. This can be done using INPVector3. You may still experience some axis crosstalk and non-linearity after the lens parameters are updated, its not unusual.

The origin point for rotation of the clip is adjusted to be around the estimated center of head rotation. This means rotating the clip around its own origin will result in x and y translation. If you want to avoid the origin point offset, you can use GetPoint and INPVector2 to get the 3DOF position of each marker and perform the rotation calculation in your own software.

(Note : we no longer provide support for INPVector/2/3 - but you can find information about it by searching these forums).

Re: Positioning and reference grid

Posted: Tue May 19, 2009 6:04 pm
by simon_o
[quote=NaturalPoint - Birch Zimmer]If you are not using a TrackIR3/TrackIR4 you may want to update the lens parameters to get more accurate readings. This can be done using INPVector3. You may still experience some axis crosstalk and non-linearity after the lens parameters are updated, its not unusual.[/quote]

Ok. I haven't found any reference on the forums to the actual values of the focal length and width/height of the TrackIR5 camera. Can you give me this information?

[quote=NaturalPoint - Birch Zimmer]The origin point for rotation of the clip is adjusted to be around the estimated center of head rotation. This means rotating the clip around its own origin will result in x and y translation. If you want to avoid the origin point offset, you can use GetPoint and INPVector2 to get the 3DOF position of each marker and perform the rotation calculation in your own software.[/quote]

Ah, I see. Now that I know this, the coordinates make much more sense.

Re: Positioning and reference grid

Posted: Wed May 20, 2009 5:22 pm
by simon_o
[quote=Simon222]Ok. I haven't found any reference on the forums to the actual values of the focal length and width/height of the TrackIR5 camera. Can you give me this information?[/quote]

the answer, in case anyone needs that, is:

TRACKIR5_PIXELWIDTH = 640;

TRACKIR5_PIXELHEIGHT = 480;

TRACKIR5_IMAGERWIDTH = 3.84;

TRACKIR5_IMAGERHEIGHT = 2.88;

TRACKIR5_FOCALLENGTH = 3.99;