Tutorials Video

How Auto-Focus Systems Actually Track a Subject

Intermediate · ~14 min

Overview

Modern autofocus isn't a single technology. It's typically a combination of a fast, direct focus-measurement method (phase or contrast detection) and, on top of that, a subject-recognition layer (increasingly AI-based) that decides what part of the frame to point that focus measurement at, and understanding both layers explains both autofocus's real strengths and its real failure modes.

What You Need

No special equipment required. This is a technical explainer, not a hands-on tutorial

Steps

The simple version: autofocus measures sharpness and adjusts the lens

At the simplest level, autofocus works by measuring how sharp (in-focus) a specific area of the image currently is, then adjusting the lens elements and re-measuring, repeating until sharpness is maximized at that area. The two classic methods for doing this measurement are phase detection and contrast detection.

Phase detection versus contrast detection

Phase detection uses dedicated sensor points that compare light arriving from slightly different angles through the lens to calculate focus distance directly and quickly, without needing to 'hunt' back and forth. Contrast detection instead measures image sharpness at the main sensor and adjusts the lens iteratively until sharpness peaks, which is generally more accurate for the point measured but historically slower since it often has to overshoot slightly to confirm the peak.

The subject-recognition layer on top: what decides where to point the focus measurement

Deciding where in the frame to point either focus-measurement method is a separate problem from the measurement itself, modern cameras increasingly use AI-based subject recognition (trained to identify human faces, eyes, or other specific subject types like animals or vehicles) to automatically select and track the correct focus point as a subject moves through the frame, rather than requiring the operator to manually select and move a focus point.

Why understanding both layers explains autofocus's real failure modes

When autofocus 'fails' by focusing on the wrong thing, it's usually a subject-recognition layer problem (misidentifying what to track, or losing tracking when a subject is obscured) rather than the underlying phase or contrast measurement being inaccurate, understanding this distinction helps troubleshoot autofocus problems correctly, since the fix for a tracking failure (adjusting subject detection settings) is different from the fix for a measurement failure (checking lens compatibility or lighting conditions).

Pro Tips

If autofocus is tracking the wrong subject, check your camera's subject-detection settings (human, eye, animal, vehicle) rather than assuming the core focus system is broken. It's usually a recognition-layer issue, not a measurement issue.
Phase detection autofocus generally performs better in continuous-tracking video situations (fast-moving subjects) than contrast detection alone, which is why most modern hybrid systems combine both.
Low light and low-contrast scenes challenge both phase and contrast detection differently, if autofocus struggles specifically in dim conditions, added light or a wider aperture often helps more directly than adjusting autofocus settings.

Knowledge Base

What You'll Learn

Modern autofocus combines a direct focus-measurement method (phase or contrast detection) with a separate AI-based subject-recognition layer that decides where to point that measurement, understanding both layers explains why autofocus succeeds or fails in specific situations.

The Plain-English Version

One system measures 'is this specific spot in focus, and which way do I need to move the lens?'. That's phase or contrast detection. A separate system decides 'what spot should I even be measuring?'. That's the AI subject-recognition layer. Most autofocus failures come from the second system, not the first.

The Misunderstanding This Clears Up

People sometimes describe a camera's autofocus as simply 'good' or 'bad' as a single quality, but the measurement technology (phase/contrast) and the subject-recognition layer are genuinely separate systems that can each succeed or fail independently: a camera can have excellent focus measurement but poor subject tracking, or vice versa.

Where This Fits

This guide covers one specific part of cinematography. The wider picture, what focal length, exposure, lighting ratio, and camera movement actually communicate, rather than only what the controls do, is in Cinematography Fundamentals: Lens, Light, and Movement, which frames the discipline as a whole and links out to the detailed guides underneath it, including this one. If you are starting from scratch rather than solving a specific problem, read that first and come back here.

FAQ

Q: Why does my camera's autofocus sometimes focus on the wrong thing?
A: This is usually a subject-recognition layer issue (the AI system misidentifying what to track, or losing the subject when obscured) rather than a problem with the underlying phase or contrast detection measurement, checking your camera's subject-detection settings is the more direct fix than assuming the core autofocus hardware is faulty.

Q: What's the difference between phase detection and contrast detection autofocus?
A: Phase detection uses dedicated sensor points to calculate focus distance directly and quickly by comparing light from different angles. Contrast detection measures image sharpness at the main sensor and adjusts iteratively until sharpness peaks, phase detection is generally faster for continuous tracking, while contrast detection can be more precise for the specific point it measures.