KM By the way, here's a graph of ambient sunlight at 300x magnification (all the other graphs on this page are 100x). You can see that it's pretty flat, so the device isn't adding much noise.
Here it is at 100x magnification. Again, virtually no noise.
I found a monitor today that has very low flicker. It's an LG 27MQ450-B. I paid $130.
My hypothesis, of course, is that I can use this monitor without headaches since the flicker is so low. It comes with a 30 day return policy, so I'll use it for a couple weeks and let you know what I think.
If it works well, I'll buy a handful of these, test them with my setup, and re-sell them at cost. Hopefully some of you will participate in this experiment.
Here are some screen shots. All screen shots are at 100x magnification.
Flicker at contrast setting of 40 and 100% brightness. This is the setting I plan to use
Same as above, except the contrast is set to 100. Interesting how the shape of the flicker changes with a higher contrast. This setting is roughly twice as bright as the previous setting.
Same as above, except the contrast is set to 0. This setting is about 1/3 as bright as when the contrast is set to 40. This is a bit too dim for me.
Also interesting: when I adjust the color settings, it changes the shape of the flicker. The default for this monitor is for R G and B to be at 50, but this screen shot has them all set to 100. Contrast is set to 40.
GregAtkinson definitely keep us updated! I need to find a safe monitor.. I'm worried polarization is also a factor for me.
GregAtkinson (1) So your eyes do not hurt from the flickering of pixels? And you can "safely" use the 6bit + FRC panel, right. (2) There are a lot of people with "photophobia" when their eyes hurt at 100% screen brightness. It is better to make tests at brightness close to zero. Radex Lupin (for $70) can do that. (3) The real 8bit AH-IPS monitor LG 27BK550Y-B has a PWM of 1% at brightness zero. Apple Cinema Display 27 (2011) is also safe for the eyes. Thank you.
Sony SDM-S205F S-IPS (2005).
100% brightness - > PWM 30%
The monitor burns out the eyes.
(in 2007, my eyes did not get tired)
Laptopmedia tests laptops from different manufacturers for flickering. Table rated (good/bad laptop). Even panel models. Why didn't I know about this site before? Add to Favorites.
AlanSmith I bought the radex lupin a few years ago. I found it helpful, but the the photodiode + oscilloscope was much more useful for me because I could see the shape of the graph, rather than just a percentage. For instance incandescent lights flicker, but it's a very smooth, gradual flicker (like a sine wave).
I do get headaches and tinnitus from the flickering of pixels, but I wouldn't say that my eyes hurt.
I can use a 6bit panel, but I generally use Intel integrated graphics which I don't think uses temporal dithering by default (I believe the default is spacial dithering, which doesn't bother me).
Bright screens bother me if they're excessively bright, but for me I don't need to turn the brightness all the way down to 0.
AlanSmith The problem with Laptopmedia is they only test for PWM, and for me, PWM is only part of the problem.
GregAtkinson I too also get tinnitus and headaches too from screens.
Which model Intel graphics ? And what 6bit screen? I have heard this from someone else too! Very interesting
jordan It's an old (about 6 years old) Dell 5567 laptop with an IPS panel running Intel HD Graphics 620. I think it's this panel: https://www.panelook.com/B156HAK01.0_AUO_15.6_LCM_overview_27890.html
I don't think it helps my headaches/tinnitus to be 6bit vs 8bit, but I suppose anything is possible.
GregAtkinson I found a monitor today that has very low flicker. It's an LG 27MQ450-B. I paid $130.
My hypothesis, of course, is that I can use this monitor without headaches since the flicker is so low. It comes with a 30 day return policy, so I'll use it for a couple weeks and let you know what I think.
While I appreciate your enthusiasm for this project there's something prudent I think you should know. Think I read somewhere that the vast majority of 1080p IPS panels are 6-bit+FRC (i.e use integrated dithering to achieve 8-bit color). You may be thinking "were there not good monitors in the past?". I think historically, 24" 1920x1200 IPS panels didn't use dithering and were often relegated for professional use, as opposed to a "consumer" product.
JTL I read a couple places that this monitor supports 167 colors. I assume that means 167 out of a possible 256. So I assume it would dither (temporal and/or spatial) to get those 89 missing colors. Another site said it uses 8 bit + FRC.
That said, my light meter can't detect any flicker caused by temporal dithering. I've tested 100+ shades of gray, red, green and blue. I'm not saying there isn't temporal dithering, but any decent* temporal dithering algorithm would be very difficult to detect with a light meter.
Personally, I believe things like backlight flicker (not just PWM), OLED refresh cycles, pixel inversion and other pixel flicker are more likely to cause headaches because those things are detectable with a light meter. I don't think temporal dithering (or spatiotemporal dithering) will cause me headaches as long as it uses a decent algorithm. The flicker is so much smaller for temporal dithering than it is for those other causes of flicker.
* I did use Ditherig to turn on temporal dithering for my laptop and on certain shades of gray the screen literally flickered from one shade to another and the flicker was visible (both with the naked eye and on my light meter). I'm assuming that most decent dithering algorithms use spatiotemporal dithering and are more intelligent about randomly spreading the use of adjacent colors through space and time.
The thorlabs light sensor is great, but could you also mention your oscilloscope and what settings you've used to measure flicker? This is very important. The sensor itself may be fast enough for PWM dimming flicker in higher frequencies, but the oscilloscope might miss half of the readouts.
Another thing to keep in mind is that light bulbs can affect your results, if you test in shops (uncontrolled environments)
More detail in regards to this topic:
https://int.rigol.com/news/blog/DS70000BLOG
If I were you, I'd consider going for a higher PPI display as well, since low ppi (like on 24" 1080p) is a known eyestrain contributor as well (regardless of distance)
Also, in regards to dithering, you might like this topic:
https://forums.blurbusters.com/viewtopic.php?f=27&t=12953
Blurbusters has a new section that tackles display-related eyestrain.
qb74 I'm using a Picoscope 2204a. All the oscilloscope settings can be found in the screen shots above. the only setting you cant see in the screen shots is the gain/zoom, which is generally 100x.
I get really clear waveforms at 100KHz. The specs say it handles 10MHz, but I haven't seen any displays (backlights) that flicker faster than 100KHz.
One nice thing about the thorlabs sensor is that it has a circular shield around the photo cell that eliminates all light from the surrounding environment when the unit is placed directly on the screen. I thought I might need to test in a dark environment, but I found that to be unnecessary.
If in doubt, any setup's accuracy can be tested by creating PWM flicker on purpose, for example with an LED connected to a Raspberry Pi: https://ledstrain.org/d/375-howto-create-pwm-led-flicker-with-a-raspberry-pi
By the way, I've had a lot of headaches/tinnitus recently so I'm waiting for a day that I don't have headaches to give the new monitor a thorough test. I used it for a couple hours on Sunday. I began with a slight headache and ended with a slight headache, which really doesn't tell us much (other than that the monitor isn't absolutely horrible).
Also, I forgot to mention that I bought the monitor at a place called MicroCenter which literally has about 50 monitors on display. I had previously gone to a couple different Best Buy locations that have about 15 monitors on display. The monitor I purchased was the only monitor I found that was 27" or larger with little to no flicker. I found a couple 24" monitors with little to no flicker. So percentage-wise, you could say that roughly 5% of monitors (3 out of about 65 or so) have little to no flicker.
GregAtkinson how pixel inversion causes flickering if the waveforms aren't perfectly symmetrical.
Thanks so much for sharing all of this!
I found the same thing with subpixel flicker asymmetry creating gross combined flicker patterns that can be detected with a meter measuring a larger screen area - data posted here: https://www.flickersense.org/testing-leds-and-screens
For me, assuming there isn’t much backlight flicker, it seems like my symptoms are worse when there’s more pixel flicker, especially when the flicker of the red, green and blue subpixels are different from each other.
I’m just skimming because I can barely look at screens anymore, so I might have missed it - does your system allow you to measure the flicker of subpixels of different colors separately? If so, have you noticed any correlation with symptoms if there’s more color-to-color flicker? For some screens that hurt me I couldn’t detect much flicker grossly with my flicker meter, but once I started quantitating the subpixel flicker with Photoshop luminosity quantitation of individual subpixels in microscope slow-motion video frames, the patterns for which screens and screen settings hurt me most started to make some sense. For example, Nightshift on my Apple devices has always been awful for causing my concussion-like symptoms and it has some extra green and a lot of extra blue subpixel flicker compared to red. When the subpixel colors flicker similarly it seems to be less harmful for me.
jen does your system allow you to measure the flicker of subpixels of different colors separately? If so, have you noticed any correlation with symptoms if there’s more color-to-color flicker?
I found something fascinating on your site: "It turns out that on every screen I've tested, Nightshift works by flickering green subpixels somewhat more and flickering blue subpixels a lot more, thus reducing the average amount of green and blue light. The overall effect is to create much more color-to-color flicker, since the screen flickers between very red and somewhat more blue/green."
I never knew why nightshift bothered me. I have lent my equipment to a friend, but I'll take a look when I get it back. I can isolate the red/green/blue subpixels using this page, which has all 256 shades of gray, red, green and blue.
The subpixel flicker mentioned on your site is what's called pixel inversion. I created this site (https://pixelinversion.com) to help visually see this flickering without a microscope. I can detect the pixel inversion flickering with the Thorlabs PDA100A2 (which has a gain/zoom feature) but I was not able to detect it with my homemade photodetector.
I see that a lot of your work has been on lightbulbs. I'm super-happy with some MaxLite bulbs that I bought that have a completely flat waveform. For testing lightbulbs with a photodetector, I've just bought a color lens filter kit so I can see the waveforms of the different colors. I'll test that as well when I get my equipment back.
GregAtkinson Thanks for the suggestion of light bulbs! I'll have to try those.
When detecting differences in the flicker of different color subpixels, for Nightshift, I measured the flicker of the different subpixels (using the microscope and slow-motion video) on a completely white (or sometimes gray) screen without Nightshift and then using the same white (or gray) screen, but with Nightshift turned on at its warmest. It's super labor-intensive to do this using the Photoshop histogram/luminosity function, but I was glad I did. Before doing this test, I could feel that Nightshift was doing something that badly hurt me, but before this test, I didn't have good evidence for why. It's dramatic how much more the green and especially blue subpixels are fligckering than the red supixels during Nightshift.
I could also detect differences in the flicker of different color subpixels (without Nightshift) using other colors on the screen. To find dramatic examples to measure with the microscope (since microscope/video measurement Photoshop analysis so labor intensive and triggering of symptoms for me), first I measured flicker with a flicker meter of different colors when only a single R, G, or B value was nonzero, at many settings. Then I chose to make 2 of the R, G, or B values nonzero, choosing combinations of values that on their own gave different flicker meter readings. Then when I recorded slow-motion microscope videos, i could see and quantify differences in luminosity of differently colored subpixels as they flickered. I could also put camera color filters between the flicker meter and the screen to get readings for different colors (Tiffen Red25, Blue47, and Green58 filters were the most useful for me). However, the microscope slow-motion video measurements tended to show color-to-color flicker best because I could measure individual subpixels and didn't have the interference between differently-phased flicker patterns that obscures readings with the flicker meter. Different screens had different colors with the most dramatic color-to-color flicker. On the screen I use most, i can mostly limit the colors that are used. I immediately feel pain/pressure in my head if I have even a small part of the screen displaying something with high color-to-color flicker.
I'll be interested to hear what you find out with your method!
