Home › Forums › Software Development › Eye position definition
Tagged: construction
- This topic has 16 replies, 8 voices, and was last updated 1 year, 2 months ago by Grant [Tobii].
- AuthorPosts
- 31/07/2014 at 21:49 #1455kangwangParticipant
Hi,
I’d like to know the exact definition of the 3D eye position, whether it means the middle point of two eye corners or the center of the pupil. Thanks a lot.
04/08/2014 at 15:44 #1466Jenny [Tobii]ParticipantHi,
It is the position of the eye globes regardless of the direction in which you are looking, so it is not the center of the pupil.
05/08/2014 at 20:06 #1472kangwangParticipantThanks for your reply, but when I try to fix my head and move eyes from left to right or from top to bottom, the eye position data (X, Y, Z) changes, and the change seems reasonable in the coordinate system where screen center is the origin, that’s why I doubt the eye position means the pupil center.
If it means the eye globe, then the eye position won’t change that much if I fix my head, so I’m confused.
06/08/2014 at 16:07 #1475Jenny [Tobii]ParticipantHi kangwang,
There will be some fluctuations of the eye position as the eye tracker’s view of the eye changes as the eye is looking in different directions, and there may be some stray points in the stream of data. Even when holding the head and eye-gaze fixed, there will be small movements in the eye position data. The data is “raw” in that sense – there is no smooth filtering of the eye position data stream. The eye position data is given in the unit millimeters but it is not millimeter perfect.
This means that the eye position data stream is mostly usable for relative positioning and averaged movements, rather than absolute positioning.
If I run the Eye Position sample in the Unity SDK (Unity 3D is downloadable for free), I can see these small fluctuations, and some stray points as I look to the far edges of the screen. But all in all, I perceive the two spheres representing my eyes in a 3D to follow my eye balls quite smoothly.
21/08/2014 at 08:41 #1520MaijaParticipantI also tried the Eye Position Stream and calculated the distance of both eyes in 3D space. When I changed the depth of the object I was looking at (I moved it at the z axis) the distance definitly changed. Thus, I got a smaller distance for objects closer to me and a greater distance for objects further away. How can that be possible if it’s the center of the eyeball which is used for the Eye Position Data?
Thanks and best regards,
Maija22/08/2014 at 10:54 #1528Mattias [Tobii]ParticipantHi,
The coordinate system is fixed at the eyetracker and therefore the screen.
The eye position is not the center of the eyeball, but better thought of as the position of the pupil. Hence when you sit still and move your eyes it will change. If you find that this does not explain the behaviour you see, then please post here and we will try to see if something is amiss.
/Mattias
28/08/2014 at 13:49 #1562MaijaParticipantThanks for your reply, Mattias.
I am still a little bit confused and just want to understand it right.
– Jenny’s answer was: “It is the position of the eye globes regardless of the direction in which you are looking, so it is not the center of the pupil”
– Your answer was: “The eye position is not the center of the eyeball, but better thought of as the position of the pupil.”Thus, the eyeposition refers to the pupil, but it does not use the exact pupil center? Is it some reflection inside the pupil you use as reference value or what do you use exactly to calculate the eye position coordinates?
Thank you for your help!
Maija
28/08/2014 at 15:21 #1565kangwangParticipantI was confused by the answers too, but from the experiments I have done as posted before, I think the eye position means the pupil, it’s position will change if you move your pupil to see different directions. Jenny said that this kind of change might due to fluctuations, but when I move pupil to look at the screen from left to right, the ‘X’ coordinate increases, so the change is following the coordinates system.
Can you help me with this? Thanks a lot.
29/08/2014 at 13:31 #1566Mattias [Tobii]ParticipantHi,
Sorry for the confusion.
The Eye position that the eyetracker hands out refers to an internal state parameter that does not translate “exactly” to a physical point, but can for all purposes be thought of, and follows the behaviour, of the centre of the pupil. i.e NOT the centre of the eyeball.
The coordinate system is righthanded and the origo of the coordinate system is at the centre of the screen with y and x axis in the plane of the screen and z pointing towards the user.
Hope this helps,
Mattias08/08/2019 at 06:56 #11853Mohammed Aatif ShahabParticipantI am still confused with the reference point of eye position. If it is the centre of screen then some values will be negative, but I have all positive values. Please clarify, How the eye position of eyes is different from Gaze points?
One more thing, If I subtract Horizontal position of the left eye and horizontal position of the right eye, will I get vergence?08/08/2019 at 16:36 #11854Grant [Tobii]KeymasterHi @asaatifshahab, and thanks for your query. Could you kindly clarify exactly which SDK, API and Tobii Eye Tracker you are using as the co-ordinate system may vary depending on the use case. That being said, the gaze point variables reflect the location on tracked screen ( normalised between (0,0) and (1,1) ) whilst the eye position is the (x,y,z) location of the each eye relative to the eye tracker.
If by vergence, you mean the average eye position between both in space then it would be sufficient to take the average value not the difference.
If you could provide further details of your setup, we will be happy to clarify. Best Wishes.17/08/2019 at 17:12 #11892Mohammed Aatif ShahabParticipantI am using a Tobii TX300 eye tracker. The coordinate system I am using is ADCSpx(origin is at the top left corner of the screen). I still didn’t understand the eye position. Please clarify the term ‘relative to eye tracker’. In my case, the eye tracker is fit at the bottom part of the screen.
My question is that when should I get a zero value for the Horizontal position of left eye? This will better help me to understand the reference point for eye position.
If possible, please give a pictorial representation.
Thanks18/08/2019 at 15:49 #11893Grant [Tobii]KeymasterHi @asaatifshahab and thanks for your query. The Tobii TX300 is part of the Tobii Pro business department.
This forum is intended for support with the Tobii Tech Consumer level eye trackers (4C, Eyex, etc) and their associated SDK’s.We would advise that you get in touch with Tobii Pro Support team directly @ http://www.tobiipro.com/contact/contact-support/ for an
answer to your query.However, that being said, I found online a reference for the ADCSpx co-ordinate system on the Tobii Pro Studio Manual (Page 136)
https://www.tobiipro.com/siteassets/tobii-pro/user-manuals/tobii-pro-studio-user-manual.pdfWhich also demonstrates pictorially the co-ordinate system in question. Hopefully this answer your query.
12/07/2023 at 11:34 #23441mae-expParticipantHello all,
I have a question: if the person is a bit mobile and moves closer to the screen for example, does this update the pupil diameter? Or is it up to us to calculate this by using eye position data?
Thank you very much
Best wishes
17/07/2023 at 14:50 #23469Grant [Tobii]KeymasterHi, the pupil size will automatically adapt to user position, so no need to calibrate according to this distance.
- AuthorPosts
- You must be logged in to reply to this topic.