delicate Y position with Eyelink2

1
Hello my name is hagen.
I work at the bauhaus university in weimar germany. we do some
research with the Eyelink2 gazetracker. We run our EM-tests with
matlab(Eyelink&Psychtoolbox) on a win2000 Pc-system and now we have
some problems with our ouput data. here a little discription. Maybe
anyone out there knows the problem and can help with a little hint.
Thanks a lot in advance hagen.

we calibrate our subjects, then present some stimuli at certain
positions. although during test, the eye positions seem to fit quite
well with the picture (on the secondmonitor), in our data files, the
y-values are wrong. they are usually(depends on calibration quality)
about 60 to 200 pixels too high, (only to high never to low).
for example: we have a good calibration. we then present a central
fixation cross (at 640, 512 for our resolution of 1280 * 1024) which
allows the next stimulus to appear only if the eyes fixate for a
certain time on the cross. the next stimulus appears (indicating that the
cross has been fixated), and the data report as coordinates for the
central fixation something like 636, 290. i.e., x is o.k., y is too
small. the phenomenon is the same for all resolutions of the display
monitor. where are the 200 pixels gone?

I am happy about every little hint. :-)
2
Also make sure that you have a similar luminance for the calibration
and the actual stimuli. Pupil size differences between calibration and
experiment might result in inacurate measurements.
Another issue is "Looking down" ! When subjects look down at a
display, the eyelid covers part of the pupil and results in
displacement of measured center of the ellips downward. Try to put
your monitor a little higher, so subjects have to look up a little
bit, and check if their eyelid actually covers their pupil when they
look at the lower edge of the display.
Best,
Arash


--- In psychtoolbox@yahoogroups.com, "franswcornelissen"
<f.w.cornelissen@...> wrote:
>
> Are you supplying the right coordinates to the operator PC about
your screen?
>
> This normally happens within initeyelinkdefaults routine, but
perhaps you change
> resolution afterwards?
>
> Gr.
> Frans
>
>
> --- In psychtoolbox@yahoogroups.com, "sojoner_84" <sojoner@g...> wrote:
> > Hello my name is hagen.
> > I work at the bauhaus university in weimar germany. we do some
> > research with the Eyelink2 gazetracker. We run our EM-tests with
> > matlab(Eyelink&Psychtoolbox) on a win2000 Pc-system and now we have
> > some problems with our ouput data. here a little discription. Maybe
> > anyone out there knows the problem and can help with a little hint.
> > Thanks a lot in advance hagen.
> >
> > we calibrate our subjects, then present some stimuli at certain
> > positions. although during test, the eye positions seem to fit quite
> > well with the picture (on the secondmonitor), in our data files, the
> > y-values are wrong. they are usually(depends on calibration quality)
> > about 60 to 200 pixels too high, (only to high never to low).
> > for example: we have a good calibration. we then present a central
> > fixation cross (at 640, 512 for our resolution of 1280 * 1024) which
> > allows the next stimulus to appear only if the eyes fixate for a
> > certain time on the cross. the next stimulus appears (indicating
that the
> > cross has been fixated), and the data report as coordinates for the
> > central fixation something like 636, 290. i.e., x is o.k., y is too
> > small. the phenomenon is the same for all resolutions of the display
> > monitor. where are the 200 pixels gone?
> >
> > I am happy about every little hint. :-)
>