From the article linked by the OP:
"That's about the time he found out about PWM dimming, a dimming method used by some LED lights, OLED displays, and other LED-based gadgets. Instead of dimming like a traditional light source, where voltage is simply reduced to produce a dimmer light output,
LEDs that use PWM dimming rapidly turn on and off several hundred times per second to fool your brain into seeing a brighter or dimmer image." [emphasis mine]
However, I question whether the bolded statement is correct, at least for Macs. Specifically, I don't think the flicker seen on Mac LCD screens is due to PWM dimming. That can be an issue on OLEDs, since it's typically in the 100's of Hz, which can be noticeable to some. But the PWM frequencies of the LED backlights on the Mac LCD panels are far higher. For instance Notebookcheck lists the PWM frequency for the 14" & 16" M1 MacBook Pros as 14,880 Hz, which shouldn't be detectable.
My guess is thus that it's instead due to FRC. From Wikipedia (
https://en.wikipedia.org/wiki/Frame_rate_control#:~:text=FRC is a form of,(half frame rate) flicker ):
"FRC is a form of
temporal dithering which cycles between different color shades with each
new frame to simulate an intermediate shade. This can create a potentially noticeable 30 Hz (half frame rate) flicker. FRC tends to be most noticeable in darker tones, while dithering appears to make the individual pixels of the LCD visible. TFT panels available in 2020 often used FRC to display
30-bit deep color or
HDR10 with 24-bit color panels."
Those who are bothered by looking at Mac LCD screens can test if this is the case by checking out the Pro Display XDR, which is a native 10-bit panel, and thus does not use FRC (at least I assume it doesn't!).
Note: By "30 bit" and "24 bit" color, they mean what's usually referring to as 10 bits and 8 bits, respectively, since they are multiplying the bit depth by 3 colors/pixel.
I’m the one mentioned in the article. I’ve learned a lot over the past year, but I don’t purport to be an expert. That being said, my personal triggers are a combination, with temporal dithering and FRC triggering severe symptoms.
The developer of Stillcolor calculated that Apple Silicon’s GPU dithering is occurring at approximately 15Hz over 4 frames per 60Hz refresh cycle. This is confirmed by the mod of r/PWM_sensitive. They calculated this by breaking the capture card footage down frame by frame. Microscope footage of Subpixels also backs this up.
The US Dept of Energy places this frequency within the “high risk” range within even 1% modulation - so it takes a very minimal amount of this to cause symptoms for many. I suspect the pattern-based nature of dithering, especially if spatial dithering is incorporated, is enough to trigger neurologically sensitive individuals. The fact that this is all occurring within 15-30Hz, well-established in epilepsy research, seems to make it risky for certain people.
I can actually use an iPhone 13 on iOS 15. It’s completely fine for me. I could not use the iPhone 13 on iOS 16 or 17, because temporal dithering was either added or the algorithm was modified. iOS 18 is more tolerable, but I can tell when the dithering is enabled because it causes tachycardia. The 2021 models of the iPhone 13 have fairly stable modulations and fairly high (as far as OLED iPhones go) PWM frequencies. It’s also a nice sine wave which helps. I can also use certain LED light bulbs with no problems. So from my perspective, it appears frequency, modulation, and waveform are key.
I can personally attest that the first (intel) 16" MacBook Pro in 2019 had AWFUL PWM dimming, and the M1 Pro (14" microLED) also had extremely bad PWM. Both devices were returned before the keyboards even got greasy.
For someone with PWM sensitivity, there's a decent but gruff reddit community where you can find the extremely limited list of non-PWM devices. FRC/Temporal dithering is really not the same, and conflating the two or mixing things up based on a lack of understanding leads to a lot of people having trouble finding a monitor they can live with.
Yes, my current MacBook is the 2022 13” M2 MBP with the Touchbar. Notebookcheck cites its PWM at 117kHz below 56%. As soon as I drop it about 50% it gets uncomfortable because the voltage change introduces noise into the signal. Notebookcheck previously detected 60Hz noise on the 13” M1 MBP, so I suspect this may be panel lottery related. Interestingly downgrading from Ventura to Monterey improved this moderately, which indicates that softest updates can alter the voltage on these devices. The screen is also not as bright on Monterey as Ventura.
This claim makes no sense.
The purpose of Pulse Width Modulation aka PWM is to simulate intermediate or "analog" amplitudes using a digital system which is only capable of generating 100% (full amplitude) or 0% (zero amplitude) outputs. Here's an explainer from Arduino documentation, with some useful example waveforms:
So long as the PWM modulation frequency is high enough to be imperceptible to its consumer, for many purposes PWM is just as good as generating a true and steady intermediate output level. But this is also why it's nonsense to talk about there being a "smoothness" or varying "amplitude" in a PWM wave. By definition, a PWM wave is always a square wave (the least smooth waveform), the on pulses are always at 100% amplitude, and the off pulses always at 0%.
Caveat: Nature abhors a true square wave, and therefore transitions between on and off in PWM take nonzero time. However, in this specific case we're talking about a drive transistor running a LED. Both of these circuit elements switch quite rapidly, so the waveform should always look reasonably like a fairly sharp-cornered square wave.
The only thing which is actually relevant to your ability to perceive PWM in LED backlights is flicker fusion. If the modulation frequency is too low, the photoreceptor cells in your eyes will be able to perceive flickering, or stroboscopic effects during rapid eye movements. But as has already been mentioned, the PWM frequency in LED-backlit MacBook LCD displays is around 15000 Hz, and there's just no plausible physical way for you to perceive flicker at such frequencies. What we know about the photoreceptor cells (rods and cones) in your eyes doesn't support it, and as a practical matter no human has ever tested to be able to perceive any flicker related effect above 1000 Hz (iirc).
One working theory among researchers is that COVID can lower a person’s critical flicker fusion threshold. I suspect this is what happened to me, because at one point prior to COVID I owned this machine and the M1 Pro 14” MBP and could tolerate them. It’s also worth pointing out that the modulation on the MiniLED MBPs is ridiculous - sometimes as high as 90%. ProMotion also causes wild fluctuations in the PWM modulation so I think there are too many triggers on those machines to narrow down a specific individual cause.
My ask for Apple is to simply offer accessibility options within settings to disable GPU dithering and TCON FRC on the MacBook panels. It’s a simple software change. Stillcolor does it with a simple command: “enableDither=No.”
PWM I know is more complicated. Apple cares about health, even offering settings for people with epilepsy to dim the screen when flashing lights appear. It’s unfortunate for some of us these computers themselves, especially with dithering, can cause seizures and migraines.
Also, as an aside, just because the eyes aren’t registering this flicker doesn’t mean the brain isn’t. That’s part of the problem. The research these technologies are based on is flawed and outdated, as far as what is safe and is not, as well as what is perceived and is not. This is by no means a settled topic within the scientific and medical community. I don’t fault companies for relying on these papers, but it’s been nearly 10 years of users bringing up these symptoms from PWM and dithering and the pandemic seems to have made more people susceptible to serious symptoms. The subreddit and online forums have grown dramatically since COVID and the amount of people I speak to who now have a problem as a result is very concerning.
It doesn’t help that both Apple and Microsoft are heavily relying on dithering as 10-bit and P3 wide color becomes the norm, and PWM is a mainstay for how AMOLED functions. These issues have to be addressed.
www.androidcentral.com
