single refresh display duration - help?

[liwenna]

Thanks Paul for that thorough reply!!

Unfortunately I did all that (and more ;)) but to no avail. The
display that follows the 13ms display is actually empty, except for a
backgroud color (gray shade). I tried using wait objects and text
display's rather than a slide display, and also the offscreen
canvasses. With offscreen canvasses the problem lies no longer with
the 'empty display that overlays the 13 ms image' arriving too late,
the timing is the fine in about 75% of the trials, but I am pretty
certain that the 13ms display is skipped in about 25% of the trials.

The graphics card might be crappy, I am also a bit suspicious
regarding the display that I was assigned for this project,
nonetheless, the 'old task' seems to be work well on it, while this
new taks also does not perform well on the pc & display that I used
for the old project (and do trust).

It simply drives me crazy that the exact same set-up in terms of
settings that are known to matter for timing, works perfectly in one
set-up (with more images and a more difficult list structure, even
with an additional procedure that has a 27ms display added to the
sequence with the 13 and 496 display) and just doesn't work in the
simpler set-up. Something must somehow matter that I completely
overlook.

I suppose I'll spent the second part of this afternoon on it again:
any additional advices are more than welcome :)

best,

liw

On 5 mrt, 21:23, Paul Groot pfc.gr... at gmail.com> wrote:
> Hi Anne-Wil,
>
> You will get the best onset performance if you:
>
> 1) set the pre-release time of the objects before both slides to the
> maximum possible value (i.e. same as duration of that object). In
> other words: if the 13ms slide is displayed immediately after the
> other slide, you must set the pre-release time of the first slide to
> 496 (or 500 if you apply point 3 below).
>
> 2) Use bitmaps instead of JPEG and make sure the color depth is the
> same as used in your exp. This might help because decoding jpeg files
> takes a lot of CPU time (although the smaller jpeg files will load
> faster from disk.)
>
> 3) use durations which end in the middle of two screen refreshes I.e,
> if the display is using a refresh rate of 75Hz, you will set the fast
> stimulus to 0.5*1000/75=7ms, and the longer stimulus to
> 37.5*1000/75=500). Not sure if this is still required with the new EP2
> release, though...
>
> Depending on the hardware performance and screen resolution, this
> still might be insufficient to get single frame stimuli. If this is
> the case, you might want to prepare the stimuli using off-screen
> canvases at the start of the sequence and copy to the display canvas
> using a fast copy at the vertical blanking interval. This should
> always work, unless the graphics card is crappy.
>
> best,
> paul
>
> 2012/3/5 liwenna <liwe... at gmail.com>:
>
> > Hi all,
>
> > I have something and I hope someone may have experienced a similar
> > thing that may help me out on this one.
>
> > I have an experiment in which a sequence of displays occurs with the
> > two most important display having durations of 496 and 13 ms
> > respectively (give or take max 2 ms). Back then, it took me quite some
> > work but in the end I got the timing right (yay).
>
> > Now... I am trying to build a second, similar but less complex, task.
> > This new is less complex since it shows only a single. rather than
> > multiple, image at the 496 ms display, and has a far less complex list
> > structure due to less different procedures. However: I can not get the
> > timing right at all..  the display that follows the 13 ms display just
> > keeps being delayed by one refresh (13 ms).
>
> > I have tried pretty much everything for hours and hours:
> > - building the new task from scratch, imitating all (or at least I
> > believe so?) settings from the old (correctly timed) task (and then
> > tweaking all the settings too)
> > - modifying the correctly timed task into the new task  (and then
> > tweaking all the settings)
> > - using off-screen canvases rather than slides (but it seems that the
> > 13ms 'canvas' is simply not shown in about 25% of the trials... which
> > is not good enough either)
>
> > I believe I have covered pretty much everything that could be done
> > about it, but I honestly fail to see why the old task would be timed
> > correctly (still is!) and any attempt at the more simple task just
> > fails in terms of timing.
>
> > I so very much hope that someone here is able to identify the one
> > thing that I just keep overlooking? >.<
>
> > Best,
>
> > Anne-Wil
>
> > --
> > You received this message because you are subscribed to the Google Groups "E-Prime" group.
> > To post to this group, send email to e-prime at googlegroups.com.
> > To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
> > For more options, visit this group athttp://groups.google.com/group/e-prime?hl=en.

-- 
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com.
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en.