How to blur faces in a video on iPhone

Can you blur faces on iPhone video? Not with the built-in tools

No, not properly. iOS ships a markup pen you can smudge over a still photo, and iMovie has no blur or pixelate effect at all — the usual workaround is a picture-in-picture overlay placed over the face. Neither one tracks a face as it moves. The markup tool only works on a single frame, and iMovie's overlay sits in a fixed position for the whole segment — it does not follow a person walking, turning, or shifting in frame.

That gap is why 'blur a face in a photo' and 'blur a face in a video' are different problems on iPhone. A photo just needs one region covered once. A video needs that region re-covered on every one of the 25 to 60 frames it moves through, and doing that by hand does not scale past a couple of seconds of footage.

Why the Photos markup and iMovie workarounds are a poor fit for video

Photos markup is built for photos. You can extract a still frame from a video, blur it with the pen tool, and save it back as an image — but that only covers one instant. Apply it to enough frames to cover a moving subject for even a few seconds and you are manually exporting, editing and re-importing dozens of separate images.

iMovie's picture-in-picture workaround covers a fixed rectangle, not a face. iMovie has no blur or mosaic effect, so you fill an overlay with a solid colour or a pre-blurred still and place it over the face. If the person is standing still and the camera is locked off, that can work. The moment either one moves, the overlay stays where you put it and the face walks out from under it. There is no mask you can drag along a moving subject, so a moving face means cutting the clip into short segments and repositioning the overlay for each one — see the full iMovie workaround.

How to blur faces in a video with Deface

Deface detects every face in a video and tracks it automatically, so there is no keyframing and no cutting the clip into pieces.

1. Import the video. Deface reads it straight from your library, up to 4K and any length.
2. Let it detect. On-device analysis finds faces and follows them as people move, turn or walk across the shot. Tracking has been improved to keep the censor aligned through motion and short detection gaps. If no faces are found, try Retry with maximum detection.
3. Turn faces on or off. Each detected face gets its own toggle, so you can blur bystanders while leaving the main subject visible, or the reverse.
4. Pick a style and export. Choose Gaussian blur, pixelate, solid colour, emoji or Eye Bar. Review how the censor follows turns and movement before exporting at the original resolution and frame rate.

Why this is a scale problem, not just a missing button

Put a number on it: a 15-second clip shot at 30 frames per second is 450 individual frames. If a face moves even slightly across that time — a head turn, a step forward, a hand-held camera drifting — a single fixed blur region covers it correctly for only a handful of those frames. Getting believable coverage by hand means either re-marking the region every few frames or accepting that the face is exposed for stretches of the clip in between. That is the actual reason there is no quick manual fix for video the way there is for a photo: it is not one edit, it is hundreds, repeated for every face in the shot.

This is also why 4K and long clips make the manual route worse rather than better. More resolution does not make tracking easier, and a longer clip is proportionally more frames to re-mark by hand — the problem scales with the footage, which is exactly the kind of repetitive, frame-accurate work a detector is suited for and a person doing it manually is not.

When this actually comes up

Street and travel footage with bystanders in the background who never agreed to be filmed. Home videos you want to post publicly but that include other people's kids. Security or doorbell clips you want to share without identifying whoever walked past. Vlogs and street interviews shot in busy places. In all of these, the subject you care about moves, which is exactly the case the built-in iOS tools were not designed for.

If you only need to cover something that is not a face — a license plate in a driveway shot, a street sign, a screen — the same manual-region tool handles it, tracked across the video the same way.