Apple · Filed Feb 17, 2026 · Published Jun 25, 2026 · verified — real USPTO data

Apple Patents an AI That Automatically Reframes Video Around Any Subject

Imagine shooting a video where the person you're filming keeps moving around the frame, and your editing app repositions the crop on every single frame so they always look centered. That's exactly what Apple is patenting here.

Apple Patent: AI Auto-Reframing for Video Editing — figure from US 2026/0181204 A1
FIG. 1A — rendered from the official USPTO publication PDF.
Publication number US 2026/0181204 A1
Applicant Apple Inc.
Filing date Feb 17, 2026
Publication date Jun 25, 2026
Inventors Paul M. Bombach, James C. Arndt, David N. Chen, Todd E. Kramer, Shaun M. Poole, Rupamay Saha, Eugene M. Walden
CPC classification 382/155
Grant likelihood Medium
Examiner CENTRAL, DOCKET (Art Unit OPAP)
Status Docketed New Case - Ready for Examination (Mar 20, 2026)
Parent application is a Continuation of 18314095 (filed 2023-05-08)
Document 20 claims

What Apple's auto-reframing video tool actually does

Picture this: you record a birthday party on your iPhone, but the birthday kid keeps wandering to the edge of the frame. Right now, fixing that in post means manually trimming each shot. Apple's patent describes a system that handles all of that automatically.

The idea is an AI-powered reframing tool that watches your video, identifies a specific subject (a person, say, or a pet), tracks where they are in each individual frame, and then crops each frame so that subject stays front and center. You tell it what to focus on, and it does the rest.

Apple trains the system by showing it pairs of original videos alongside already-edited versions, so it learns what a good reframe looks like. Once trained, it can apply that knowledge to any new video you hand it, without you touching a timeline.

How the ML model learns to crop and follow a subject

The patent describes a two-part system: a training phase and an application phase.

In the training phase, Apple feeds the machine learning model large sets of video pairs, each consisting of an original wide shot and a corresponding edited version where someone has already cropped the frames around a subject. The model learns the patterns behind those human editing decisions.

  • Subject identification: For each frame, the system pinpoints where the target subject is located.
  • Selective cropping: It selects the portion of the frame that includes the subject and discards the rest.
  • Frame-by-frame consistency: Because the subject's position is re-detected in every individual frame, the crop adjusts continuously as the subject moves.

The system also accepts a set of modification rules (essentially user instructions, like "keep this person in frame") that guide which subject to follow and how to handle the crop. The end result is a fully reframed video generated without manual editing, ready to watch as a finished clip.

What this means for iPhone video editing

For most iPhone users, the gap between a decent recording and a watchable video is the editing step, and that step is where most people give up. A tool that automatically keeps your subject in frame closes that gap in a practical way, especially for the vertical-to-horizontal (or horizontal-to-vertical) format conversions that come up constantly when people shoot for one platform and then post to another.

Apple already offers a version of auto-reframing in Final Cut Pro and iMovie, but this patent points to a more AI-driven, trainable approach rather than simple rules-based tracking. If this technology makes it into consumer apps, it could make the "multi-cam" and auto-crop features available to people who would never open a professional editor.

Editorial take

This is a real, practical improvement to a problem anyone who edits phone video has hit. It's not a flashy concept patent, it's the kind of infrastructure work that ends up improving iMovie or the Photos app in a future iOS update. Worth paying attention to if you care about where Apple's video tools are going.

Get one Big Tech patent every Sunday

Plain English, intelligent commentary, no hype. Free.

Source. Full patent text and figures from the official USPTO publication PDF.

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