OLED questions from a visual neuroscientist

[moregabor]

Dear Forum,

I am a university researcher in neuroscience who would like to ask the community here about some luminance behaviors of OLEDs, in particular the ASUS PG27AQDM. We are currently trying to find a suitable replacement monitor for laboratory experiments on human visual perception. While the field has long used old-fashioned CRTs due to their great response times, the newest generation of 240+ Hz OLEDs like the ASUS looks interesting. However, the monitor shows some odd luminance behaviors that we do not quite understand. Maybe the experts in this forum can help out?

Question 1: Dynamic reduction in luminance for small stimuli in the absence of ABL

In our tests, we of course find that monitor decreases its luminance when bright stimuli increase in size (Auto Brightness Limiting, ABL due to changes in Average Picture Level, APL). We have therefore carefully reduced the "Brightness" settings of the monitor to a level where ABL does not happen anymore. Surprisingly, however, we still observe in our photodiode measurements that when we present a small bright stimulus (size about 2% or 5% of screen area), then this stimulus will dim by about 5% within the first second. This dimming does not happen for larger (e.g. full-screen) stimuli, even though they are shown with the same luminance (because there is no ABL anymore). This same behavior was seemingly also observed by gaming publications. It happens regardless of whether the "Uniform Brightness" mode is activated.

Does anyone know why this happens? Does this phenomenon have a name?

Question 2: What causes the monitor's 240 Hz flicker?
A more basic question: The ASUS OLED flickers at 240 Hz. Can someone point me to sources explaining what exactly causes this flicker in an OLED monitor? I assume it is related to the line-by-line builing up of the image but would appreciate any technical insights of where this flicker exactly comes from in the absence of a backlight and how it changes according to the luminance of the displayed stimuli (it seems to be largest at intermediate grey levels)?

Thanks!
MoreGabors

[thatoneguy]

That monitor operates in sync with the refresh rate so for 240hz it flickers every 4.16ms(1/240hz), if you have it at 60hz it would be 16.67ms and so on.

asustftcentral.png (3.99 KiB) Viewed 886 times

asusrtings.jpg (129.4 KiB) Viewed 886 times

https://tftcentral.co.uk/reviews/asus-r ... d-pg27aqdm
https://www.rtings.com/monitor/reviews/ ... d-pg27aqdm

[moregabor]

Thank you! Our measurements look similar. Are there any sources on how exactly this 240 Hz flicker is generated? The tests that you have linked unfortunately do not seem to comment on that.

Regarding Question 2, does anyone know what causes small but not large stimuli to dim within the first half second even under conditions where the monitor does not show Auto Brightness Limiting? The following photodiode recording illustrates what I mean, you see the 240 Hz flicker, but you can also see how luminance decreases over time after showing something small and bright on the screen (here, a small white rectangle). Unfortunately it looks like I am not allowed to post or link an image of our results here. Any insights are highly appreciated.

[Chief Blur Buster]

It's my understanding on the thin film transitor active matrix system (found on nearly all lithographed LCD and OLED screens) -- the gate of OLED transistors have a very brief change of voltage as the pixel refresh hits the transitor. In a row-column addressing system (grid of subpixels), you are refreshing essential one pixel at a time, usually left-to-right, top-to-bottom. The screen may be subdivided to many zones (like vertical strips) to allow concurrent refresh, depending on how the row-column addressing is designed. You can see high speed videos of scanout at www.blurbusters.com/scanout

Since OLED pixels consume a lot more power than an LCD pixel (Being a direct light source), it is particularly sensitive to transistor gate voltage changes. There's a extremely tiny voltage fluctuation as the gate voltage is "reset" to the color (Even for the same color). There is also a fluctuation for LCDs but the LCDs require a lot less power per pixel (a lot less wattage), and the LCD pixel response is much slower, so it's sometimes below the noisefloor for LCDs (imbalance in voltage inversion is more visible in oscilloscope measurements).

Also, what rationale are you researching flicker? Sometimes flicker has led to a red herring effect (e.g. it's not consistently always the person' cause of eyestrain). Other factors such as antiglare discomfort versus glossy discomfort, to others can apply here. Keep in mind an even a mere incandescent light bulb has a ~5% flickerdepth at 120 cycles per second (two zero-crossing events for 60Hz sinewave AC), from filament cooling effects.



There are over 100 ergonomic issues of screens, so it's easy to be redherring'd (though flicker IS a common cause of eyestrain, it must be acknowledged as an Error Margins / Potential Assumptions, that it is not always necessarily the flicker).

Now... this is generally invisible compared to macro things like a full black frame insertion. As an interesting demo of black frame insertion on 240Hz OLED, see TestUFO Variable-Persistence Black Frame Insertion for 240Hz, which demonstrates display-persistence motion blur is proportional to pixel visibility time -- (also useful demo to see: www.testufo.com/eyetracking#speed=-1 -- watch the 2nd UFO for 30 seconds, and observe where low sample and hold rates vibrate like a slow music string, and high sample and hold rates blurs like a fast music string). I welcome researchers to write papers about these kinds of effect, and cite the TestUFO tests (like many other papers did). OLED is otherwise a very pure "sample and hold" display.

[thatoneguy]

To add to what Chief said: There is also the factor of most LCDs using DC Dimming at a very high rate(some high-end models can do 50khz).
DC Dimming is pretty unfeasible with OLEDs since it would affect color accuracy and so on and so forth, so most OLED manufacturers use PWM or sync to refresh rate.

[moregabor]

Wow, thank you for these extremely helpful responses! I now have a much better understanding of the source of the flicker and useful pointers to additional sources.

I should clarify that we are not specifically researching the effects of flicker; instead, we are trying to understand current OLED technology better to see whether the newest generation of 240+ Hz OLEDs can serve as a true CRT replacements in research on vision and cognitive (neuro)science due to their excellent response times.

Is there any way that fairly new Forum members like me can post a link or image (this is currently blocked for me)?
It would be great to get you opinion on our plots showing the dynamic reduction of luminance mentioned in my Question 1, which oddly only occurs for small, but not for large or full-screen stimuli. This change in luminance is not huge (~5%), but since stimuli in my field need to be tightly controlled for their exact luminance (e.g., to measure a person's perceptual thresholds for faint stimuli), it may be a problem for the research use of OLEDs. So it would be interesting to better understand where this behavior is coming from and whether it can be prevented.

[Chief Blur Buster]

Is there any way that fairly new Forum members like me can post a link or image (this is currently blocked for me)?

You should be able to post images now.

One comment.

For saying "true CRT replacement", it most assurdedly certainly is not, as CRT is an impulse driven display and OLED is a sample and hold display.

However, if you meant "true CRT replacement for the purposes of the tests we need to do", then perhaps. Colors, average luminances, etc. It's (usually) a much better CRT replacement than LCD in those attributes. Now, if you're concerned about motionblur or latencies, then it is not as good at CRT in those specific lineitems.

It depends on the test. Tests that depend on motion blur or depend on latency, will typically be affected. However, the extremely high refresh rate of the OLED will greatly compensate. Keep in mind that not all pixels refresh at the same time -- see www.blurbusters.com/scanout -- this is true for both CRT and OLED. There's the scanout latency, and there's the tapedelay latency (totally independent latency interferences).

I simply say this, as a citable error margin, for research papers. Understanding what limitations you hit for specific tests you do, will be key in determining if it's a "true CRT replacement" for the purposes of specific tests.

No display is currently (yet) a total all-checkboxes-checked CRT replacement.

Anyway, if you enjoy Blur Busters articles, explore www.blurbusters.com/area51

[moregabor]

Yes, with "true CRT replacement", I was only referring to the use in my research subfield as a stimulation device for evoked brain response recordings (where input lags, response times, and true blacks are critical).

Below you see a plot of the dimming phenomenon from my first question: We notice that small but not large bright stimuli on the ASUS OLED dim rapidly within the first 500 ms by about 5% (or 10 nits).



As the upper plots shows, this does not happen for large stimuli (> 25% of screen area). Importantly, at the luminance level at which these measurements were taken (Brightness setting: 40%), ABL does not happen anymore, that is, the "sustained" luminances (measured after 1 sec or so) are unaffected by how big a stimulus is. I would have expected such a "dynamic dimming" (is there an official name for this?) maybe for large bright stimuli, but not exclusively for small bright stimuli.

A 5% change in luminance for some stimuli (only small ones) is not necessarily a deal-breaker for research use in my field, but it would be helpful to understand where this is coming from, what it depends on, and whether this phenomenon has a name or a technical explanation.

[thatoneguy]

Your image is not showing.

I believe you meant to post this.

oleddimmingpuzzle.jpg (116.27 KiB) Viewed 279 times

PS: You should preferrably use attachments to post images, since linking from dropbox seems to vanish the image after a while.

[thatoneguy]

Also, some other tidbits you might find useful:

OLEDs have no phosphor decay so no phosphor trail artefacts during dark scenes like with CRTs.
Overall the pixel response time which is in the microseconds is not as fast as the electron gun on a CRT(nanoseconds), so from that perspective they aren't a match for CRT and we will have to wait for MicroLED or Emissive Quantum Dot.

However from a motion clarity perspective the upcoming 480hz OLED monitors come close to a FW900, and a 1000hz OLED would be as good as most consumer PC CRT monitors.