Disable dithering on M1/M2 devices

nilanganray

Nope. Only thing was I had disabled true tone. But enabled it for weeks now and still have symptoms. Coincidence or a late onset?

Are there any solutions to this? I tried to do the dummy display thing but my mac freezed totally.

async

nilanganray Try to disable night shift and true tone. That helps me at least. If you switched wallpaper, switch back.

nilanganray

async

There's one thing. The laptop comes with a two-pin cable so there's a grounding issue where I live. The first month of usage I was mostly using it unplugged and the second month, I have mostly been using plugged. I am wondering if this is causing an issue or not.

async

nilanganray Could be low-power mode. It supposedly reduces ProMotion and HDR content. Doesn't seem like people can agree upon exactly what it does. You can force it even when plugged in under settings > battery. There is some option for power saving in Chrome as well that might be involved. Setting refresh rate to 60hz and color profile to srgb would probably achieve mostly the same thing.

I've also witnessed cases where a computer won't turn on or a monitor runs too hot to use the controls from being plugged into an extension cord with other devices, or from unstable / weak power supply. No idea if it in theory can affect the monitor tho.

There used to be a bug on Macbooks would make them run hot when the power cord is plugged in on the right side of the machine as well, as the power would travel across the entire machine. It usually showed as kernel_task consuming a ton of cpu, which is Apple's way of preventing anything from using cpu to reduce the heat. It took me a few years before I saw that and figured out why my old machine was lagging.

nilanganray

async Not using low power mode. Also this is Macbook M2 Air. So no Promotion.

The issues started roughly during the time I started using it on power all day instead of power. And, the enclosure tingles because of 2 pin and grounding here. I am not sure if correlation equal to causation here.

async

After seeing this thing I stopped ruling out random theories https://mastodon.social/@haeckerfelix/110272427676278609

async

nilanganray I meant that if that was the turning point, forcing it to low power mode might make it better. It does it automatically by default when switching between battery / power, so you have to switch it if you want the same mode while connected.

The enclosure tingles? And it stops if you pull out the power cord? Make sure you don't have your feet on something grounded like wall mounted panel heater or lamp.

async

I tried digging around to reverse engineer.

@NewDwarf What method did you use to locate the potential flag? I tried digging around, but my experience with Swift and C is limited.

With simple string search there were numerous references to dithering, but it is hard to know what is actually in use, and what is some type of old driver or sdk.

Surely there must be some patterns to look for to find hidden bootflags. There are a ton of knowledgeable people in the Hackintosh community as well that might be able to give pointers as to how to figure out different flags.

Reverse engineer Mac binaries and kernel extension. https://www.hopperapp.com/
Search for strings inside binaries. https://github.com/excitedplus1s/jtool2
Docs for jtool2. http://www.newosxbook.com/tools/jtool.html
Dyld cache extractor for driver not present in /System/Library and /Library https://github.com/keith/dyld-shared-cache-extractor/
Output a ton of display information. Could be used to diff after changes. Maybe worthwhile to compare the diff to what is present in safe mode. https://github.com/joevt/AllRez
Not tested. https://github.com/0x36/ghidra_kernelcache

async

I did a quick test with BetterDisplay EDID export and AW EDID Editor on M1 and my external monitor. Forcing the bitdepth to 6 or 8 bits works fine and causes proper banding on shadows and other things.

Also one thing that hit me. If people put sRGB as the color profile and leave on true tone or night shift then it would most likely shift the colors away from the ones displayable by the original sRGB color space, and force temporal dithering. The same might apply to cell phones as well, as most of the web uses old sRGB colors for most things. However switching the EDID should most likely not have this issue, as the OS won't believe it can send those types of colors at all. Not sure if the 8 bit affects the entire stack, or if color table modifications or anything can end up being applied on the top by the gpu or monitor itself forcing it to use the dithered ones. Most likely contrast adjustments on the monitor would.

caboy

async

Is this verified to disable dithering? Some of us have seen even with the color bit depth set down via other mechanisms, the dithering is still present.

async

If someone is sure a specific update is causing issues, it would be interesting to run a comparison on the output from AllRez from before and after the update. As well as checking which kexts updated.

https://github.com/joevt/AllRez

DannyD2

Hi everyone, I'm new to the forum 🙂 I just wanted to say thanks and share that the "sudo nvram boot-args="enableDither=0 PE_i_can_has_debugger=1 iomfb_RuntimeProperty_enableDither=0" command seems to work for me on my MBA M2 15-inch on Ventura. It makes a huge difference in the specific area that was causing me eyestrain -- I can't thank you enough.

I did have to disable SIP though. About that, I have a question… is it necessary to add the "PE_i_can_has_debugger=1" flag? Can I leave it out?

Also, assuming that flag is necessary, can I re-enable SIP partially with "csrutil enable --without nvram --without debug" and keep the effect?

ryans

DannyD2 I did have to disable SIP though.

What happens if you don't disable it? There's an error printed somewhere? Or there's no impact to your eyes?

Any chance you want to test this with VideoDiff and a lossless capture card?

DannyD2

ryans

If SIP is enabled, I get the error message: Error setting variable - 'boot-args': (iokit/common) not permitted.

async

You can run this as well to check if enableDither changed

ioreg -l -d0 -w 0 -r -c AppleCLCD2 | grep -i -B 8 -A 0 dither

karthi3219

async I am not sure Dithering is disabled or not. Can you please help me on this?

I have followed these steps.

Restart in recovery mode
run: csrutil disable
Restart in normal mode and in terminal run:
sudo nvram boot-args="enableDither=0 PE_i_can_has_debugger=1 iomfb_RuntimeProperty_enableDither=0"

restart in recovery and set: csrutil enable
restart your machine

After all the above steps done, I have ran your given command.

ioreg -l -d0 -w 0 -r -c AppleCLCD2 | grep -i -B 8 -A 0 dither

Now output in the terminal is below.

Is Dithering disabled now? In the result, it still says "enableDither" = YesI am not sure this is correct and I have missed something.

DannyD2

async

Thanks for this. I see "enableDither" = Yes, so it doesn't seem like it worked after all.

caboy

karthi3219

From my reading of the output, I would guess the line that says "enableDither" = "yes" would mean dithering is enabled.

How does it feel to your eyes?

Have you done a video slo-mo test? (This may or may not be helpful without a microscope.)

karthi3219

caboy i feel the eye still hurts. That means the commands not working for my laptop? Macbook m1 pro Ventura.

Hacklord

Hi All, I have 16" MacBook Pro with M1 Max chip. I've been watching this topic since the first message, and also experimenting myself a lot. As for the nvram boot arguments, I've run myself into thinking that not all reverse engineered parameters may have the same name as their corresponding boot args, so I added few more variants. Finally, now I have the following (pure placebo maybe) opinion to have all of the following boot arguments: boot-args="-v intcoproc_unrestricted=1 kext-dev-mode=1 amfi_allow_any_signature=1 amfi_unrestrict_task_for_pid=1 PE_i_can_has_debugger=1 cs_enforcement_disable=1 amfi_get_out_of_my_way=1 amfi=0xff cs_debug=1 ipc_control_port_options=0 PE_i_can_has_debugger=1 iomfb_RuntimeProperty_enableDither=0 dither=0 enableDither=0 iomfb_force_block_Dither=1 iomfb_RuntimeProperty_force_block_Dither=1 iomfb_enableDither=0 enableDither=0" (and also I have csrutil disabled and spctl globally disabled, and of course no auto-brightness and no true tone).

But, what's true, in IORegistry the "enableDither" CLCD2 property doesn't change. To try to change the value in run time, we need to write a program (ext) which will change the property from kernel space, one way is to use the following call defined by Apple in their IOKit.h header: kern_return_t IOConnectSetCFProperty(io_connect_t connect, CFStringRef propertyName, CFTypeRef property). Maybe Amulet Hotkey thing did the same trick, they had an ext which used to work for older Intel machines and it really worked.

MTNEYE

Hacklord interesting. Have you had any further luck with this? Thanks for your work

DannyD2

Hacklord

Do you know if anyone is working on a kernel extension that uses IOConnectSetCFProperty() to set this value? It seems like it wouldn't be that complicated.

ascourtas

Hacklord do you know if anyone has figured out the "enableDither" CLCD2 property? I have an M1 Macbook Pro, and I've tried both the BetterDisplay dummy mirroring approach and the args setting approach and neither has worked for me. Notably, I too get "enableDither = Yes"

I can tell it hasn't worked because I have an e-ink monitor, and the temporal dithering is apparent as a horrible "shimmering" or movement on the monitor. My 2019 Intel Mackbook Pro does not have this issue.

« Previous Page