I disabled dithering on Apple silicon + Introducing Stillcolor macOS M1/M2/M3

async No rebooting did nothing. I tried toggling everything I could. Luckily I have a Mac Mini M1 working atm which I'd migrated the air setup to before I installed Adobe Media encoder. I thought the Mini did not work for me but it was on 10bit color mode so changing that to 8bit seemed to fix it. It's annoying some apps trigger something for me.

async Have you checked this?
https://github.com/belvederef/visual-snow-relief-overlay/releases/tag/v1.6.1
It is not that heavy on the CPU, but the most annoying thing for me is that it is seen as a "space" in Mission Control.
It helped me in the past, but it's not a true solution. If you find a way to incorporate your overlay, let me know.

anon123 I'll check it out. I have visual snow, but it's pretty mild. Noise pattern definitely does something. The plan is to package the different overlays in some app where others can experiment as well.

I've tried a lot of things and I'm sure there are modifications that can be done that either provide relief, or assist in some sort of adaptation. I usually use a shadow around the edge of the screen and feel that it helps quite a bit. I've also experimented with what changes prevents visual distortions from happening, as it is easily testable to see when patterns are blinking and not. Diagonal patterns can also be used to remove botching thar is present on Mac.

Tbh after spending a lot of time with overlays on scaled resolutions on Mac I'm getting the feel that it's the closest you can get to living inside a compressed jpg hellhole. Just try the inverted framebuffer in BetterDisplay if anyone wants to see just how much shit there is. Currently I only do retina or native scaling, but need to get a 220 ppi monitor to make it better.

Apart from this there is a lot of things that can give pixel inversion flicker. Especially when not using scaled resolutions that smooths out things. Overlays can mitigate some of this.

I do need to find a way to do more advanced overlays tho, as there is no reason you couldn't do shaders that doesn't shade images, or just affects pure whites. Mening you could make everything feel like paper without breaking images, or just pattern all white backgrounds.

I also think there is benefit to be gained from shifting whites and blacks differently. For example adjusting the gamma curve so blacks are more blue and whites are more red, or applying different patterns on them.

I've been pondering a bit around adaptation after seeing this where they have shown adaptations for months after minutes of exposure. https://michaelbach.de/ot/col-McCollough/

This is also super interesting in relation to high gamut colors on high contrast displays. At a certain threshold you get the depth effect that might be contributing. They also note something about the size of pixels and the wavelengths for it to happen. Might be one more clue as to why certain screens work better. https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2015.00337/full

async Just try the inverted framebuffer in BetterDisplay if anyone wants to see just how much shit there is

Strangely, I've noticed that using the inverted grayscale framebuffer option (specifically the BetterDisplay one, not the normal Mac Invert Colors option) helps reduce strain somewhat on M1 Air when running at a scaled resolution like 1440x900 (yes, I do mean the extremely "compressed JPEG" ones you mentioned). This is probably because -- although all of the post-processing and sharpening is still there -- everything including the post-processing (and even the black bars in non-native aspect ratio resolutions) gets inverted and at least some of it is converted to monochrome which probably reduces the effects of some of the processing. It doesn't "fix" the M1 Air, and I still vastly prefer the 2016 13" MBP which is very usable for me, but it definitely reduces my strain by a small but noticeable amount. The non-monochrome inverted framebuffer also seems to improve something too but the grayscale one works even better.

After using the framebuffer invert, you can use an invert colors browser extension to make your browser look "normal". BTW this is also the closest you can get to the iOS "double invert" technique on a Mac. This will actually highlight the compression artifacts even more -- this directly and visually exposes what the Mac is "secretly" doing to the screen.

Interesting semi-related find that I can't test at the moment: search "scaler" in IORegistryExplorer, and within one of the results, you can actually find a property called "DisableDeringing" that for some reason is one of the few display-related flags that's elsewhere instead of within the usual IOMobileFramebuffer flags. Not sure what this does but I wonder if it affects the artifacts in some way.

DisplaysShouldNotBeTVs This is super interesting. I'll review everything at home.

I posted some things about the scaler at some point. Searched a lot for somewhere to config it. I think things like surface compression, max framebuffer layers etc are involved in when it decides to do excessive compression.

One thing I really would like to see is for someone with a capture card to capture the same screen with both a screenshot and with the capture card.

async Yes, I use a low level API for that. There are actually several APIs to manage Presets, including a high level one that Apple made for System Settings - it's a bit clumsy but otherwise works fine. If you are interested, I can send you the details, but you'll need to do heavy coding to utilize it.

FNP7 Thank you! Many of the techniques should work on iOS/iPadOS as well, but did not experiment with it.

Donux Yes, you can use terminal commands. Download betterdisplaycli or just use the binary inside the app bundle. Example:

betterdisplaycli set -n=displayname -quantization=0.5

Here is more help: https://github.com/waydabber/BetterDisplay/wiki/Integration-features,-CLI

async I was planning to add some shader effects to BetterDisplay. Right now if Metal overlays are enabled in the app (a bit difficult to make that happen now as you need to change some awkward settings) you can do some simple color adjustments with it only (temperature, RGB control etc) - I primarily added this to support a night shift like effect on non-native displays (on which gamma table adjustments don't work) but did not expand the featureset much. So much to do…

async

Here is more proof that Intel is interested in the false 3D effect

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6126933/

"Computer-graphics engineers and vision scientists often wish to create display images that reproduce such depth-dependent blur, but their methods are incorrect. […] We found that accommodation is not driven at all by conventional rendering methods, but that it is driven surprisingly quickly and accurately by our method with defocus and chromatic aberration incorporated."

Funded by NSF Research Grant BCS-1734677 and by Corporate University Research, Intel Labs

Here are a bunch of highlights from the above article

"Example stimuli generated using our rendering method. The left, middle, and right images are simulated to be farther, at, and nearer than current focus"

"LCA were as effective at driving accommodation as real changes in focal distance"

"our rendering technique provides a useful signal for accommodation"

"spherical aberration can in principle provide directional information to guide accommodation"

"We are currently developing methods that enable real-time updating and are measuring how effectively they can be used to drive accommodation and create realistic depth appearance"

"The color effects associated with LCA are generally not perceived […] accommodation and depth perception results are good evidence that LCA contributes to visual function even though the resulting color fringes are not necessarily consciously perceived."

(^ this is not true LOL I can see the color fringing so easily)

This paper is specifically about creating the effect by modifying existing 2D images

"In an earlier paper, we describe how to do this for complex 3D scenes using computer graphics (Cholewiak et al., 2017). Here we describe the general problem of which blur kernels are most appropriate for creating realistic retinal images. And then we describe a method for implementing realistic blur for 2 D scenes"

They know that this can cause depth conflicts

"Our experiment presented a cue conflict in which LCA indicated that a response was required while the other cues indicated that none was needed."

They know this can simulate something floating in front of the screen

"When we simulate that the object is nearer than current focus (positive defocus), blue is displayed sharper than green which is displayed sharper than red. Our results and those of Cholewiak et al. (2017) show that presenting such a stimulus causes the viewer's eye to accommodate nearer such that it becomes focused in front of the screen"

They know that it creates a blurry image, and that people can notice this

"Instead of maximizing image sharpness as defined by those metrics, accommodative responses to simulated changes in focal distance (specifically, Defocus + LCA) actually reduce sharpness"

"an accommodative response in the Defocus + LCA condition necessarily causes more blur in the retinal image. Subjects actually noticed this, but nonetheless made responses in the direction specified by Defocus + LCA"

DisplaysShouldNotBeTVs is there a way to turn this off? Is it mentioned anywhere when this stuff was added to drivers?

Just curious does anyone know if nvidia or amd does this sort of thing ?

async I advice everyone to actually look up some chromostereopsis image on an OLED screen and see how extreme the 3D effect is with oversaturated colors

The really interesting part is that the chromostereopsis images even "temporarily" affect typically safe displays, which is what really nailed this down for me as possibly THE core issue.

For example, my 2016 Xiaomi Redmi 3 with LineageOS 18.1, which is one of the most comfortable mobile devices I own, typically looks very flat and is 100% usable for me.

(The only phone I have with a screen that looks even flatter is the 2014 Nokia X on Android 4.3, which is sooooo comfortable but too slow to really do anything with aside from typing notes)

However, the moment I put certain chromeostereopsis patterns on the Redmi 3 screen, the false 3D effect suddenly appears out of nowhere, and the pattern creates the same exact "vibrating and shaking" + "strain when trying to focus" feeling that I get from modern devices, especially on the red areas of the pattern.

Side note:

The first time I EVER experienced a device with a false 3D effect (and the corresponding eyestrain and brain fog) was in 2015 — it was the GPD XD Android gaming handheld, it ran Android 4.4, although had more obscure specs such as a Rockchip CPU.

The moment I started playing games on it I noticed that the game "looked 3D" immediately which confused me so much. It was the first time I ever felt "depth" outside of a true stereoscopic 3D screen. However I kind of brushed it off initially, because I had no idea back then if it was from the screen or if it was "just me".

Although my other devices at the time like my old laptop, luckily, were (and remain) safe — even then, something was very off about that console (but I wasn't yet aware that it was connected to the 3D effect).

I had difficulty playing games and following what was going on, and sometimes I felt a light "seasickness" after playing. I was otherwise skilled at video games back then, so I wondered why I never completed any games on that system. Now I know why.

I also vividly remember noticing red artifacts/color fringing next to all white elements on the handheld's screen, which 100% lines up with my current theory.

Edit 9/10/2024: I was incorrect that 2015 was the "first time" I experienced it. The first device I had the depth perception issue was a Late 2008 MacBook Pro I bought in 2012 (although to a milder degree)… I still have that laptop today and it still has that issue, on both macOS and Linux, it just clicked in my memory that I actually do remember it looking like that back then. I have evidence.. one image I found in the downloads folder from 2014, I actually remember saving because it "looked 3D" on that screen

Weridly, a 2009 MacBook Pro I just bought recently (that actually has more PWM than the 2008) is surprisingly fine in Linux, unlike the 2008.