Poor accuracy

[janeborodat]

I’d like to ask if the overall accuracy of the tracker wil be improved with the Eye Tribe tracker software updates, because right now I find it rather poor, even after excellent calibration.
Here is the video where it can be seen how cursor systematically is shifted to the sides of the dot where I’m looking at.
https://drive.google.com/file/d/0B6p-JZ ... sp=sharing
What can be done with that?

Thank you.

[Martin]

Hi Jane,

There's a couple of factors that influence the accuracy of the calibration. First thing to check is the physical setup, not too far from the device, placed just below the display etc. Bright ambient light might cause issues (e.g. close to windows etc). Also, the screen size should not be larger than 24".

To get a better accuracy it's possible to calibrate with more points. You will see a better result with 12 or 16.

[aamanieu]

i also have a precision problem using the mouse cursor simulation: i am off by a inch horizontaly (to the left) and an inch vertically( to the top). my setup is a 14 in ch laptop, at a distance of about 50cm from the sensor (sensor 5mm below the screen). this is about 3degree accuracy error we are very far from the 0.5-1degree accuracy advertised. how can i improve my accuracy ? thank you.

[aamanieu]

the fundamental problem is that it works perfectly if you stand still but accuracy gets really poor (an inch off in x and y axis on a 14" screeen)when you lean forward or backward (even just by a few inches) so it is not usable in real world, because users always lean forward and backward when standing still on their chair.

has this BIG problem been fixed since version 0931 ?
thank you

[JeffKang]

the fundamental problem is that it works perfectly if you stand still but accuracy gets really poor (an inch off in x and y axis on a 14" screeen)when you lean forward or backward (even just by a few inches)

I think the following thread attempts to address the problem:

Distance from sensor incurs vertical shift (and how to fix) - viewtopic.php?f=9&t=32

Here's a video showing the result (Video is unlisted) http://www.youtube.com/watch?v=0PvtbHz_arE

On its way. This feature will mark version 1.0. It's not just a simple compensation coefficient, it's a whole new calibration model.
There's a couple of things left to fix and we won't ship this feature until it works really well. Preliminary results are pretty spectacular.

[aamanieu]

insteadof using your limited dev reesources to port version 0.9 on mac why not fixing your core technology and working on delivering version 1.0 for windows first ?

[Martin]

Good point, but it is not the same people doing those tasks. It takes a wide range of competences to develop these systems. I've seen your 5 other posts about the Z-axis movement causing vertical offset and understand that it is a big limitation for you. All I can say is that we are working on it, please be patient.

[ChristopheG]

Martin,

Statements like "we are working on it, please be patient" reflect rather poorly on the product and the company. Could you give us a sense whether v1.0 is days, weeks or months away?

Thanks

Christophe.