Stream Engine Coordinate Systems

[Axel Williot]

Hi
I use the Stream Engine with the 4C and I am confused about the coordinate systems.

When relating to the coordinate systems described in http://developer.tobiipro.com/commonconcepts/coordinatesystems.html, is it correct that
– tobii_gaze_point_subscribe gives the gaze position in the ADCS?
– tobii_gaze_origin_subscribe gives the eye position in the UCS?
– tobii_eye_position_normalized_subscribe gives the eye position in the TCBS?
– tobii_get_track_box gives the coordinate transform between UCS and TCBS?
– tobii_head_pose_subscribe gives the head in the UCS?

The API doc refers to “millimeters from the center of the display”, that does not seem to match to the definition of the UCS, which is centered on the tracker, so what system is it exactly?
(When I test, it seems the eye position Y is not centered on the center of the display, but lower, maybe on the tracker. But I might be wrong.)

[Grant [Tobii]]

Hi @axelwlt, thanks for your query.

– tobii_gaze_point_subscribe gives the gaze position in the ADCS? Yes
– tobii_gaze_origin_subscribe gives the eye position in the UCS? No, it is given in the same units as UCS but with the center of screen as its origon, with the Z-axis perpendicular to the screen.
– tobii_eye_position_normalized_subscribe gives the eye position in the TCBS? Yes
– tobii_get_track_box gives the coordinate transform between UCS and TCBS? Yes
– tobii_head_pose_subscribe gives the head in the UCS? No, same as tobii_gaze_origin_subscribe. See the documentation for the rotation.

Please let us know if we can be of any further assistance. Best Wishes.

[cockcomputer]

hi Grant,
i have the same problem with Axel Williot,when i use tobii_gaze_origin_subscribe API to get the eye position,The accuracy seems not very well,especially for position Y,position X is centered on the center of the display,but position Y is lower than the center of the display and higher than the center of the tracker about 50 millimeters lower,position Z has about 100 millimeters deviation。As i update the lastest core software,this problem still exists.
i want to know why this happen and how to solve this issue.look forward your answer sincerely.
thanks.

[Grant [Tobii]]

Hi @cockcomputer, unfortunately there is not a great deal one can do to improve accuracy of reported eye position beyond the following environmental considerations:

Ensure eye tracker is not inline with direct sunlight or other strong sources of Infrared radiation (such as open windows, spotlights, etc), Reduce ambien lighting in the computer environment if possible. The presence of thick glasses, heavy make-up or facial pearcings can also interfere.

After running the user calibration, how does the accuracy test look for you (ie the 9 circle test). Also the gaze overlay feature included in the Tobii Core software might be useful to see if there are any inherent issues in parts of the screen.

You might also try on a another user to see if the deviations are persistent.

[cockcomputer]

Hi Grant,
thanks for your answer,i use the same traker 4c device for 32 inch screen and previously used is 27 inch in the same place and with the same laptop ,the accuracy of Y position would be improved to 10 millimeters divation,while Z position divation has not been imporved,
and if i change another user(without glasses),after running the user calibration,the result has not been changed.
what confused me is that why the result is very different between 32 inch screen and 27 inch screen with the same tracker 4c after calibration.and 32 inch screen seems better than 27 inch.

thanks.

[Grant [Tobii]]

Hi @cockcomputer, certainly we do not expect a significant difference in accuracy between monitors of 27″ and 32″… may I ask if the environmental conditions are the same for both monitors? Could you kindly also tell us how you are determining the accuracy in these metrics? Thanks.

[Author]

Posts