Sony PVM-A250 innards [cognitive science research]

[mike.lwrnc]

Hi folks,

I work in a cognitive science research lab. We can't tolerate the transition-variable pixel settling times that come from an LCD (i.e. white-to-black can take a few ms, but 25%grey-to-75%grey can take tens of ms), so we have a Sony PVM-A250 OLED monitor. Using a photocell and labjack, I validated that it has fast and consistent transitions. The monitor appears to only offer a 60Hz refresh rate and there is PWM such that, under default operation, the pixels are on for the first 8(ish, will round here-on in) ms, and off for the last 8ms. I say "under default operation" because there is a "flicker-free" mode under which the pixels are on for 4ms, off for 4ms, on again for 4ms (same frame content), and off again for 4ms.

Some of my work does not involve moving images, just sudden appearances and disappearances, and in these cases (to make subsequent mathematical modelling simpler), we'd prefer to have no PWM, leaving the pixels on for the entire refresh. I also wondered if it might be possible to get the display to go to 120Hz (somewhat implied by the display's ability to do a double presentation of the same frame under flicker-free mode), so I contacted Sony to see if any firmware or hardware tweaks might be possible. However, in a subsequent conversation with a product support guy, it was noted that modification of the PWM and implementation of 120Hz was counter to the intent of the display.

With no official support for such modifications, and with pretty much nil expertise on my own to even evaluate whether they are possible, I thought I'd post some shots of the display's innards here in case anyone is interested and can provide input. Here they are: http://imgur.com/6P279L6,hurNawz,3b4RFZY

[flood]

Some of my work does not involve moving images, just sudden appearances and disappearances

from my understanding, those displays, like most displays, are rolling scan. i.e. if you display a white frame after a black frame, what you see in a high speed video is that the top of the display is updated first, and the bottom is updated after about 16ms (for 60hz)

the same is true for most lcds and all crts

the exception is lcds with strobed backlighting (what we call lightboost, blur reduction, ulmb, etc... here). for these the pixels are updated from top to bottom, but the backlight only turns on once the bottom pixels have finished updating. so a white frame appears as a single flash. as the backlight is off during the rest of the cycle, the pixel transition time is effectively hidden for the most part

this may be more suitable for what you need, but i dont know whether the strobing would be okay for you

[spacediver]

in our lab, we use the Viewpixx 3D which has a scanning backlight (I do a lot of psychophysics with stimuli that require precise timing, and we also use motion stimuli). http://vpixx.com/products/tools-for-vis ... ewpixx-3d/

You might also find fig 19 in this paper somewhat relevant: http://www.journalofvision.org/content/13/7/6.full

No idea how to go about manually modifying the OLED you have to increase the refresh rate, or to change it to full sample and hold, although there are some experienced tinkerers here on the forum.

Not sure what kind of modeling you're doing, but the visual system is likely integrating luminance over a 16.7 ms period, so it might not matter too much whether it's strobed or sample and hold, or double pulsed as it seems to be in your case.

[Chief Blur Buster]

Have you tested the suitability of recent 120Hz TN monitors for stimuli?

Based on my tests, a modern "1ms" strobe-capable 120Hz TN monitor will have all transitions (GtG of all possible combinations) mostly effectively >90% complete by the first 2ms. There will be lag and ghosting for the remainder of pixel transitions, often for several milliseconds, but the purchase of a 120Hz monitor, even for 60Hz stimuli, even with its imperections, might fall within your error tolerances. I'm not sure what vision stimuli you are testing, so it definitely can be problem, or it might not be --

On modern 120Hz gaming monitors, 3D-capable and strobe-capable pixel overdrive intentionally optimizes GtG to be as quick and efficient as possible because pixel transitions must be complete in the sync interval between refreshes, for 3D stereoscopic purposes and blur-reduction strobing (high speed video of lightboost). Often a portion of the optimizations done for overdrive, is kept intact in non-strobe mode, leading to very fast "mostly complete" GtG even for 25% and 75% transitions (even if a bit slower than white/black, but at least substantially 90%+ complete). 3D-capable stereo LCDs had to optimize all GtGs to be as fast as possible, due to the nature of sequential frame presentations that wipe as much as possible, of the previous refresh cycle. They are not perfect, but are substantially better than pre-2012 LCDs. You could also use a photodiode to verify the completeness of TN pixel transitions.

Also, I presume you're already familiar with the top-to-bottom scanout behavior, even of PWM-free LCDs, where the top edge of the screen refreshes before the bottom edge? This is true even for sample-and-hold displays, even OLED. You are aware of vision stimuli behavioral differences between sequential-presentation (e.g. LCD and CRT) and global-presentation (e.g. plasma) displays?

Also, eliminating PWM brings back motion blur -- you are already aware of the sample-and-hold blur effect at http://www.testufo.com/eyetracking ? Even with instant pixel response, there will always be motion blur due to the sample-and-hold effect -- see Why Do Some OLEDs Have Motion Blur?. However, if this is not a concern or error margin in your vision tests (e.g. testing only static images, etc).

If you want an OLED and one that's got no PWM, there's a model of OLED HDTV that doesn't use PWM (when its blur reduction is turned off). I'm not sure if it's the LG or Samsung, but this could be an option, if you want to avoid the flicker of the Sony PVM display. But that also means you're going to have to live with a larger screen size, e.g. 55"

Perhaps you are also studying the persistence-related-blur effect and you're looking for a "mathematically perfect sample-and-hold display"? Such a display would follow exactly the mathematics of 1ms of persistence equalling 1 pixel of motion blurring per 1000 pixels/second motion (the formula which I used to invent/verify the checkerboard optical illusion at http://www.testufo.com/eyetracking#pattern=checkerboard .... The more mathematically perfect the squarewave sample-and-hold effect is on the display (frames displayed for full refresh cycle and cleared quickly next refresh cycle) the stronger the checkerboard optical illusion becomes. It's more muddy on older IPS monitors, while very crisp on most 120Hz+ TN monitors.