How Does Your Phone Know It's You?

Описание: How does Face ID work on iPhone, even in the dark and through sunglasses? Every time you glance at your phone, it throws a net of invisible infrared dots across your face (Apple put the count at more than 30,000), reads the shape of your face from where they land, and checks it inside a sealed room in the chip.

We follow one glance: the flood illuminator, the dot projector and the infrared camera, a depth map built from the dots, why many sunglasses let infrared through, the Secure Enclave and the Neural Engine, how a neural network turns a face into a list of numbers, and why a photograph can't fool it while twins and children sometimes can.

WATCH NEXT
How Did Phone Cameras Get This Good?    • How Did Phone Cameras Get This Good?  
What Happens Inside an AI Chip When You Ask a Question    • What Happens Inside an AI Chip When You As...  
How Does AI Actually Know Anything?    • How Does AI Actually Know Anything?  
Playlist, How Everyday Tech Works:    • How Everyday Tech Works  

CHAPTERS
0:00 How does your phone know it's you?
0:54 Light you can't see: infrared
2:14 A net of dots: the dot projector
4:55 Through your sunglasses
6:00 Down into the chip: the Secure Enclave
7:58 What it sees: your face as numbers
10:33 Tricksters: photos, masks and twins
11:59 It grows with you
12:35 One yes
13:08 Credits

CREDITS
All animations and illustrations in this film are original works of Silicon Up Close, made in Blender from Blender Studio's free Human Base Meshes (CC0): no real person's face is shown in them. The dots, the lasers, the depth landscape, the chip drawn as a city and the map of faces are our illustrations, not the design or layout of any product; infrared, which our eyes can't see, is drawn in red. The phone in them is a plain one with no maker's mark.
Footage from Pexels (Pexels License): Engin Akyurt, Rafael Martinez, cottonbro studio, K, Orhan Pergel, Florian Delée, KATRIN BOLOVTSOVA, MART PRODUCTION, German Korb, Julia M Cameron, Artem Podrez.
NASA's "Black Marble" (2012), the Earth at night: NASA Earth Observatory, via Wikimedia Commons (public domain).
Thumbnail: conceptual AI-generated artwork; the man is not a real person.

RESEARCH
Apple: "About Face ID advanced technology" (support.apple.com/102381); Apple Platform Security ("Biometric security", "Facial matching security", "The Secure Enclave"); "If Face ID isn't working on your iPhone or iPad Pro"; "About Touch ID advanced security technology"; Apple Newsroom, "The future is here: iPhone X" (12 September 2017), Finisar and VCSELs (13 December 2017), iPhone 15 Pro (12 September 2023).
Patents Apple owns, from PrimeSense: US 8,749,796 and US 8,150,142 (they describe methods, not the phone).
Pyo, Kim, Oh, Mah and Jang, "Evaluation of Optical and Thermal Properties of NIR-Blocking Ophthalmic Lenses Under Controlled Conditions", Sensors 25(11):3556 (2025).
Schroff, Kalenichenko and Philbin, "FaceNet: A Unified Embedding for Face Recognition and Clustering", CVPR 2015 (our map of faces pictures that published idea, not Apple's network, which Apple doesn't publish).
WIRED (Andy Greenberg), 14 November 2017: the boy on Staten Island. Coors Field: MLB.com (50,144). San Francisco: US Census Bureau, 1 July 2025 estimate (826,079).
For iPhone 18 Pro owners: Apple says these models also use extra image data, handled in a separate part of the chip it calls the Secure Exclave. This film follows the system Apple has described since 2017.

Our sums: about 8.2 billion people doing one sum a second would need about 73 seconds to match one second of 600 billion operations; a billion pictures at one a second is about 31.7 years.

#FaceID #iPhone #HowItWorks