Design Problems With Face ID on the Mac

Oct 25, 2021 • Extra Ordinary

People have been asking for Face ID on the Mac ever since its debut on the iPhone X in 2017. I’m not surprised it still has yet to come to the Mac, even after every Mac in the lineup has since been redesigned. There are three design problems with Face ID that Apple has yet to solve for the Mac.

The dot projector

According to the iPhone X announcement keynote, Face ID doesn’t just use a camera image of your face. It also uses a flood illuminator, infrared sensor and, crucially, a dot projector to make a 3D map of your face.

The dot projector uses “over 30,000 dots.” Think of it like a 3D image with a resolution of 160 × 200—that’s 32,000 pixels. For reference, this is what a portrait at 160 × 200 looks like, taken at the distance I hold my phone from my face:

Now, compare that to a picture taken from the point of view of my Mac at my desk, cropped to match the size of my face in the image above:

The dot projector becomes significantly less precise at a farther distance. This is why Face ID only works within about 18″ your face. It’s good enough because your phone is designed to be held in your hand. The iPhone doesn’t need to support the wide variety of distances and angles people sit at their desks, but a Mac will.

Apple could simply train their neural stack to account for this with no change to the hardware, but I do not believe they would accept an implementation of Face ID that is half as accurate as the iPhone.

The display panel

The display of the MacBook Pro is thinner than every iOS device by a huge margin. In fact, since the first retina MacBook Pro, there is no separate display panel inside the lid; the aluminum lid is the back panel of the display. The iPad Pro is 0.23 inches thin compared to the MacBook Pro display at 0.13 inches—that’s 43% thinner. While I could not find the exact depth of the Face ID sensor array component, according to this picture from the iFixit teardown of the iPhone 11, the component appears to be too thick to fit inside the MacBook Pro display.

This component doesn’t just have to be thinner than the 0.13 inch-thick-display, it has to fit between the glass and the aluminum of the display.

Authentication consent

The dot projector accuracy and component size are both hardware problems that could, in theory, be overcome with intense engineering resources to design a more robust sensor array in a thinner package. There is still one issue that is a matter of design: consent.

You must place your finger on the Touch ID sensor to authenticate with Touch ID every time. Authentication and consent are the same step; you place your finger on the sensor, it checks if it is your finger.

Making sure your face looks like your face is largely an involuntary action, so Face ID needs an additional step to make sure you consent. Otherwise, a malicious app could ask the system for access to the password keychain or other sensitive files, wait for a Face ID dialog and instantly accept with no input from the user.

Would an on-screen button or a standard system dialog that accepts hitting the Return key be secure enough? On the iPhone, after Face ID has successfully authenticated, Apple asks for your consent by asking the user to double-click the one button developers cannot control: the side button. They could very well do this on the Mac by adding a special button to the keyboard that only the Secure Enclave can access—but does that sound familiar? We would arrive at a system that works exactly like Touch ID but with more expensive engineering and parts.

This doesn’t mean Face ID won’t happen

These are not impossible design challenges to overcome. In fact, I bet Apple already has a big, ugly prototype Mac with Face ID working in a lab somewhere in an Apple Park basement.

Researching how to make Face ID work on the Mac will make Face ID better and more secure on the iPhone by proxy—smaller components make more room for bigger camera hardware and a smaller sensor area. If or when Face ID ships on a Mac is a matter of improving the technology, miniaturizing it and then engineering a method of manufacturing the components without adding a significant cost to the product.

Good thing the only company that has the ability to do this happens to be the richest company in the world.