I find it unfortunate that m1 macs have this problem. This person's creativity could have been put to use on something other than fixing a flaw in a computer.

Yeah, I made this app as I switched a family member's mac from an Intel mini to M1 and was mystified by the fact that no matter what tricks I was trying, it was not working with the Lenovo 24" QHD display as before (getting a 4K display was not an option due to cost and size constraints - no good 24" 4K display is available for a reasonable price and there ppl who don't want huge displays on their desks). I purchased a HDMI dummy but had constant issues with it, so I had to explore other avenues.

My creativity is normally put to use on something other, like MonitorControl, but working on BetterDummy is fun as well. :)

https://github.com/MonitorControl/MonitorControl

Probably doesn’t qualify as reasonably priced, but this 24” monitor has suited me well. I shrink everything by one notch and everything is very crisp and “retina”.

LG UltraFine 4K Display

https://store.apple.com/xc/product/HMUA2VC/A

Now that is software I have a use case for! I used some abandoned menu bar app for this. Nothing worked, except the manual slider, which was enough for me.

I guess I'll have to try this with my display!

What does it actually do? What is the point of a fake display that doesn't exist?

It's useful on a headless Mac used a server.

Is this a joke or a serious answer? Why would a headless mac need one more display that it doesn’t use?

When you remote into the Mac with VNC (which is the protocol Apples native Screen Sharing solution uses) you are limited to the display resolutions that the hardware thinks is available. This tricks the hardware into allowing additional resolutions that you otherwise wouldn’t be able to get over a VNC session.

This is just a completely foreign concept for Windows users. My main work PC is a laptop with a 4k built-in display, and 2x 1920x1080 external monitors. I can RDP in from my home machine, which has a single 5120x1440 monitor, with zero issues.

Only if you use RDP which is basically a remote graphic card where as VNC type systems capture the frame buffer of actual graphics card on the remote. Steamlink, Team Viewer, etc all have similar issues on Windows as well.

Tried doing Steam link to a Windows computer without monitor attached and would only work in like 1024x768, needed a monitor attached to get 1920x1080 almost bought a dummy for the Windows machine similar to this but had and extra monitor so I didn't bother.

Also VNC like system will turn on a monitor and you can see what people are doing remotely, RDP doesn't do that. RDP is a really nice system that gets really good performance over low bandwidth links due to remoting higher level graphics API, at least it used to, wish Apple had a built in equivalent that worked similarly.

Oh yeah, I know that RDP is superior. But Apple's own first-party remote desktop solution is basically VNC, and suffers from similar limitations. They could be more like Microsoft, but they seemingly don't care.

NoMachine is a great alternative for RDP. My remote desktop only had Win Home which have the RDP features disabled, so NoMachine is an answer to that. There is a command line to enable the RDP in WinHome but it require several step and external library to make it functioning. Now I have a new Intel NUC with WinPro, so I ditched NoMachine for MS RDP. NoMachine works flawlessly on macOS and Windows (and it have Linux support). I would say NoMachine is trailing behind MS RDP while Apple VNC is closer to the totem pole.

MacOS still uses the garbage VNC protocol instead of what Windows did, which was treat remote display as a first-class product; indeed RDP is one of Windows greatest features that is unheralded/underappreciated.

Haha this is even crazier than the original fake-mirroring usage!

I'm guessing to reap benefits of https://en.m.wikipedia.org/wiki/Supersampling

On Windows, I use AMD's VSR (Virtual Super Resolution) to set 5k screen resolution on a physical 4k display, letting me set the UI scale factor to 200% and have reasonable font sizes with clean icons (125% and 150% looks fuzzy). Under Linux, it is an xrandr spell that achieves the same thing.

How does this work? Don't you have to scale the virtual 5k resolution down to 4k? Wouldn't that also generate fuzziness?

If a stroke is 1px wide at 100%, 2px at 200% / 5k, how does that stay sharp when you scale it down to 80%?

Or is the idea that whatever algorithm AMD uses is better than the one Windows uses?

Say you have a 28" 4K display but you want to run 2560x1440 resolution to get you desired UI size. You can either run 2560x1440 scaled up to 4K, or 5120x2880 scaled down to 4K. The latter will look better.

Or you run at 4K and let applications scale – which sadly not all support properly.

This is what I usually do. I run 4k at "100%", but then I increase font size as needed. I usually don't care about the icons, widgets, etc since I rarely interact with them, so to me, it's actually a good thing they're small. This way everything is plenty sharp, and I also get more screen real estate.

The reason I use a 4k screen is to have sharp text, so I try to avoid scaling.

That’s why – despite being a free software advocate – I currently run Windows for all critical software, as on Windows I can scale content without re-scaling windows. (Qt and Android support it as well, but Gnome and macOS don’t).

I'm not using Gnome, but scaling the text works fine for GTK apps on I3.

Yes, but Elementary, Android, Windows and Qt can scale icons and all elements together with the text to any arbitrary scale, pixel-perfect, without any post-processing.

GTK can only scale text, or scale the whole window to 2x, 3x, etc and then scale it back down in post-processing.

Downscaling doesn't bother me as much for some reason. Icons scaled to 125% or 150% from their 1x size look more blurry than the 2x version of the same icon scaled down. It also looks better to me if the system uses integer scaling to blow up the 1x icon to 200% size first, then scale down from there.

The github page describes the exact scenario that this is supposed to solve. I don't think I can explain it any better than that page already did.

Is this the "explanation"?

> To fix this problem, BetterDummy creates a virtual dummy display which you can then utilize as a mirror main.

What is a "mirror main"?

By using display mirroring (i.e. having all your displays show the same output) you can effectively use scaling settings determined applicable for the primary display that the OS wouldn't normally allow you to apply to the secondary displays. This is a workaround solution for that OS limitation in which a fake primary display is created to give you more settings for the real display (by making the real display the secondary).

Thanks, that's actually an explanation. Seems crazy that there's no simpler solution!