Finder 工具栏图标在非视网膜显示器上呈锯齿状或消除锯齿

Question

I'm trying to create Finder toolbar icons that match the look and feel of Monterey, but they seem to be jagged as if there are issues with anti-aliasing.

Here are the retina icons I'm trying to convert (two on the right):

And here is how they look on my non-retina display:

At first, I thought this was simply a problem with automatic anti-aliasing. However, when I manually resize the icons to 32x32, this is how they're supposed to look:

But when I use those icons in Finder, they look jagged all over again. I have no idea what is happening.

Update

Steps to reproduce:

1. Take icon_terminal_light.icns from this icon pack
2. If you already have a custom Finder toolbar icon, skip to step 6
3. Right-click on your Finder toolbar and choose "Customize Toolbar"
4. Click-and-drag any application onto the toolbar
5. Click "Done"
6. Right-click on your application and choose "Get Info"
7. Click-and-drag icon_terminal_light.icns onto the icon in the upper-left

Notice:

If you move the Finder toolbar to a non-retina screen, the icon will look thin and the edges will be jagged.

You can even open icon_terminal_light.icns in Preview, resize it to 32x32, and then export it to a new icns file. The icns file will look smooth on a non-retina monitor, but if you try putting it into the toolbar, it'll be jagged once again.

Answer 1

Finder has a quirk whereby the alpha layers of toolbar icons get multiplied by two when the window is active. If the Finder window is in the background, however, the icons probably look how'd you expect.

If your icons have any thin graphical element, the anti-aliased pixels will have low alpha values. When those alpha values are doubled, the pixels practically disappear, which is what looks like is happening.

I solved this issue when trying to resize the icons provided in OpenInTerminal-Light . My solution is a bit of a hack:

To fix the rendering, I manually resized the icons and added 89% gray behind the anti-aliased pixels. Every pixel has either 0% or 100% alpha now.

You can see more detailed notes in the issue I opened on GitHub .