Google Patents a Way to Judge Your Distance From a Camera by Reading Your Iris

By Patentlyze Team · Updated Jun 12, 2026

Every human iris is roughly the same physical size — and Google wants to exploit that biological constant to figure out exactly how far your face is from a camera, no depth sensor required.

How Google measures eye distance using your iris

Here's a quirk of human anatomy that engineers love: the colored part of your eye — the iris — is almost the same diameter in every adult, roughly 11–12 millimeters. That means if a camera can measure how many pixels your iris takes up in a photo, it can work backwards to calculate how far away your eye actually is.

That's the core idea in this Google patent. The system maps a mesh of tiny reference points across your face and eyes, uses those to measure your iris size in the image, and then runs a calculation — factoring in the camera's focal length — to estimate the distance between your eye and the lens.

Once it knows that distance, it can modify the image accordingly. Think of it like giving any ordinary camera a basic sense of depth, even if the hardware has no dedicated depth sensor. Portrait blurring, augmented-reality overlays, or beauty filters that behave correctly in 3D all become more precise when the system knows how far your face really is.

How iris pixel size maps to real-world depth

The patent describes a pipeline that starts with a standard 2D image and extracts depth information from a single biological measurement: iris pixel size.

• Facial mesh construction: A machine-learning model identifies dozens of landmarks across the face and eyes, assembling a mesh that includes precise eye and iris markers.
• Iris pixel dimension estimation: Using the iris-specific landmarks, the system calculates the iris's apparent size in pixels within the image frame.
• Depth calculation: The system uses a classic optics equation — combining the iris pixel size, the camera's focal length in pixels (a measure of how strongly the lens magnifies a scene), and the offset between the iris center and the image's optical center — to estimate real-world eye distance.
• Image modification: With that depth estimate in hand, the system can alter the image — adjusting blur, repositioning AR elements, or applying effects that depend on knowing where in 3D space your face sits.

The key insight is using a mean iris dimension — a population-level average of real iris size — as a known physical anchor. Because this anchor is biological and nearly universal, no dedicated depth hardware (like a LiDAR chip or structured-light projector) is needed. A plain RGB camera is enough.

What iris-based depth means for AR and portrait modes

Dedicated depth sensors — the kind found in higher-end phones or AR headsets — add cost, power draw, and hardware complexity. A system that can extract reliable depth from a single camera image is attractive for cheap devices, web cameras, or any context where you only have one lens to work with. Portrait-mode blurring, face-unlock systems, and AR filters that need to position objects in 3D space all get more accurate when depth is estimated well.

For Google specifically, this fits neatly into its existing MediaPipe face-tracking work — the same open framework already used by developers building real-time face effects on Android and the web. A depth layer built on iris size could upgrade millions of apps that already use Google's face-mesh tooling without requiring any new camera hardware from you.

Editorial commentary on a publicly published patent application. Not legal advice.