Problem with Pygaze OS plugins using xpyriment backend
Hi all and happy new year!
kicking it off, I am having problems using the xpyriment backend for my eyetracking experiment, using Tobii tx300 eyetracker.
first, the dimensions in which the calibration is presented in the screen as well as the entire experiment display in the screen seems off.
I am using a 1024 x 768 display, and while the pygame backend displays both the calibration and the experiment in those dimensions, when using the xpyriment backend the calibration dots don't move all the way to the coordinates of these coordinates, they occupy a much smaller rectangle (if you were tio draw one) and the same is true for the instructions in the experiment, even though the stimuli per se occupies the entire screen (1024x768) as it does with pygame. Also, the calibration only runs once, the validation is not shown, nor the results of the calibration, but I can see in the debug window that the calibration is usually unsuccessful when using this backend, while in pygame things go fine. As a result, my log files in xpyriment don't have any data, other than just the log messages. I can't quite get why is this happening.
A second issue is that when I look at the debug window while the experiment is running using the expyriment backend I get these warning] canvas.show() took 54.0 ms repeatedly.
Any ideas why this might be happening, as I would like to use xpyriment backend for timing purposes, and psycho backend is not working with my stimuli.
On another note, has anyone here experienced reduced sensitivity using tobii TX300, such that it affects you calibration or takes a while to find an ideal position for gaze tracking?
Thank you!
Helio
this is on windows 7, OS 3.2.5
Comments
Hi Helio,
Thanks for the elaborate description! Just to get my head around the issue: The PyGame back-end is working perfectly, but you're unsure about its timing accuracy, correct? How come your stimuli are not compatible with PsychoPy?
As for your issue: I haven't seen many issues with timing in the PyGame back-end, but obviously this depends in part the specifics of your setup (both hardware and software). You can at the very least check whether your screen timing is consistent with the expected timing considering your framerate. This will indicate whether vsync is active, and properly engaged. (Although, to be perfectly sure, you'd ideally check for tearing with a high-speed camera, and check your timing with a photodiode.)
The delay in the Expyriment back-end seems back-end specific, and I would not be able to tell you why it occurs. Unfortunately, I simply do not know enough about how expyriment is implemented.
Regarding your TX300 question: I have seen this, but by no means in any systematic way, and this was using the old SDK. So this is not worth much
Cheers,
Edwin
hi @Edwin thanks for the reply!
Yes the pygame is working fine, when I did the math, the timing was fine, the difference between the pygaze message logs and the screen time is negligible I think (about 2 ms). So I guess it's not a big deal.
Regarding the PsychoPy backend, during the time where the videos are supposed to be displayed, I simply get a grey screen. I have tried converting my stimuli to a different format, and changing the display properties, and nothing. I get this grey screen as if something was playing, but you can't really see anything.
Tomorrow I will be testing on a different machine and a different tx300, just to see if this might be related to my current setup.
cheers,
Helio
Hi Helio,
It's good to know that at least the PyGame back-end works! I would advise using that.
Sorry I couldn't be more helpful in hunting down the precise bugs in the other back-ends. You could try flagging them as Issues on the GitHub page for OpenSesame: https://github.com/smathot/opensesame/issues In that way, they won't be forgotten about, and at some point someone with more expyriment expertise will be able to look into them
(Won't help you now, but will at least help future users who might have the same video issue!)
Cheers,
Edwin
Thanks @Edwin,
will do that!
Cheers,
Helio