Re: LCD question

1
>Hi Denis,
>... In relation to an experiment
>we're trying to run here, I've been browsing the PsychToolbox eGroup and saw
>several references to LCD projector screens used in fMRI. Are you familiar
>with the mechanics of LCD screens? I was under the impression that they
>have
>slow response times (usually 25ms or slower), with the response time being
>defined as the time it takes a cell to turn from black to white and to
>return
>to black again. I haven't found many LCD screens that claim to have quicker
>response times than 25ms, yet most fMRI projectors, including the one we use
>here, claim ability to display frames at 15ms. From my understanding, the
>time
>it takes a cell to change in an LCD screen will vary according to the
>degree it
>has to change (e.g., longer to change from white to black than from white to
>slightly less black). It seems that worst case, an image could be displayed
>and erased at only 25ms and depending on the pixel values in the image,
>different parts of the image would be on the screen longer than other
>parts.
>If you've explored precise stimulus presentation with LCD screens, maybe you
>could direct me to a good reference. Thanks
>Brian

dear brian

i only have a bit to add to what you already know. there have been
reports of slow LCDs. However, I suspect that they are mostly out of
date, as it is now popular to use the new digital input LCD screens (eg
on my powerbook g4) to show DVD movies. I do not see evidence of trailing
on these screens: moving objects look sharp. this suggests to me that
these new models are faster than what was measured and reported in the
past. Note that the LCD industry is very hot right now with very fast
technical advance.

i think there's no substitute for measurement of the actual device you're
considering. Take a photo of a white moving bar on a black field. Or
measure the temporal profile of a step change with a fast photometer .

i would strongly suggest avoiding an LCD with analog input (ie most of
what's sold today). they have internal electronics that sample and
digitize and can introduce very tricky aliasing effects and full-frame
delays. The new thing is digital input, either through DVI input, or
built-in LCD display of laptops.

the digital input LCDs driven at native hardware resolution seem to pipe
the information directly to the pixel. that's what you'll want for
producing experimental stimuli.

You may want to look at this chapter, which touches on the topic briefly:
Brainard, D. H., Pelli, D.G., and Robson, T. (2001). Display
Characterization. To appear in the Encyclopedia of Imaging Science and
Technology. Wiley.
http://color.psych.ucsb.edu/brainard/characterize.pdf

best

denis

ps
i'm cc-ing to the psychtoolbox forum since many people are asking about
this topic lately.
http://groups.yahoo.com/group/psychtoolbox/messages/
2
dear tim

good points and questions.

>On 7/7/01 11:23 AM, "Denis Pelli" <denis.pelli@...> wrote:
>> i only have a bit to add to what you already know. there have been
>> reports of slow LCDs. However, I suspect that they are mostly out of
>> date, as it is now popular to use the new digital input LCD screens (eg
>> on my powerbook g4) to show DVD movies. I do not see evidence of trailing
>> on these screens: moving objects look sharp.
>
>Not my experience. on my TiBk, if I run a smooth pursuit trial string
>(copied below), I get distinct snail-trails which are about 30 msecs long
>(at, say, 6dB reduction in amplitude).

well that's the right test to make. i defer to your results. thanks for
the report.

>Thanks for that Denis. One question I have: with the new pure-digital LCD
>displays from Apple - what is the meaning of frame rate? Is it the case that
>these are now like perfect vector scopes where the state of each byte of
>video-RAM is quasi instantaneously reflected on the screen? If not, what, I
>wonder, determines how often each pixel looks at video RAM?

I don't know. I suspect that at present everything is still serial, in
raster scan order. I think DVI is set up that way. But, as you say, the
future could be random access. Peter Lennie got the impression, from
talking to apple engineers, that this is on their minds.

>I have a feeling that the TiBk is analog (to allow for the VGA type analog
>-out connector) the block diagram suggests this also
<http://developer.apple.com/techpubs/hardware/Developer_Notes/Macintosh_CPU
s-G4/PowerBook_G4/index.html>

the evidence is very circumstantial, but my guess is still that the
pathway to the screen is all digital. the block diagram is vague. the VGA
output port is indeed analog, but recall that it carries the signal for
an independent image, unrelated to what's being displayed on the built-in
LCD. from a design point of view, it seems to me that Apple would want to
eliminate the cost, space, and power associated with having to add an ADC
and digital frame store inside the display.

perhaps, if we think about it more, we might come up with some sort of
test that would be diagnostic. One issue is whether there is a frame
delay between sending the pixel and affecting emitted light. Another is
whether the resampling (if it occurs) has measurable quantization or
aliasing effects. Tests that assessed these three possible effects would
be helpful, even if they don't definitively answer the "is it digital?"
question, because ultimately what we most want is simply to characterize
our rendering.

incidentally, we've set up some simple tests for characterizing CRT
monitors, targetting their weaknesses, and it would be very nice to add
some new tests that target LCD weaknesses.
http://psychtoolbox.org/tips/displaytest.html

best

denis