To blur a single object rather than the whole background, you need an editor that can select that object on its own. On iPhone this is done by tapping the thing on the canvas: the phone segments the frame into separate foreground objects and picks the one under your finger, then confines the blur to that shape. It re-derives the shape from the live frame every few frames, so it survives the object turning and reshaping, but it is not a motion tracker and a subject who walks across frame needs a tracked overlay instead.
There is a real gap between the two things people mean by "blur it out".
One is depth: soften everything behind the subject so the face pops. Every phone camera does a version of that. The other is targeted: blur that thing, the license plate, the whiteboard, the other person in shot, and leave the rest of the picture alone. Those are different jobs, and the second one is the one most editors handle badly.
Why the usual workarounds are bad
The two things people reach for both fail for the same reason: they do not know what the object is.
A blur rectangle is a shape you draw and hope stays over the thing. It has hard corners that scream "something was hidden here", it covers more than it needs to, and the moment the object shifts, the rectangle is covering the wrong pixels. Every frame you get wrong is a frame someone can pause on.
Background blur goes the other way. It blurs everything that is not the main subject, which is fine for a talking head and useless when the thing you need to hide is standing right next to you, in the foreground, clearly part of the scene.
What you actually want is a selection shaped like the object.
How tap-to-select works
Modern iPhones can do this on device. Apple's Vision framework has a request that segments a frame into its salient foreground objects and hands each one back as a separately addressable instance mask, available from iOS 17 onward.
The useful part is separately addressable. The frame does not come back as "subject and background". It comes back as several objects you can pick between. So the gesture becomes: tap the thing you want, and the editor keeps the instance whose shape covers the point you touched.
That is a better phone gesture than drawing a box, and it produces a mask with the object's real outline rather than a rectangle's.
What Zella does with the selection
In Zella on iPhone you tap the object on the canvas and choose one of six treatments:
- Blur object for the privacy case. The softening is confined to the object's shape, so nothing outside it changes.
- Spotlight, which darkens everything else instead. Use it when you want to draw the eye rather than hide something.
- Outline and Glow, the two "look at this" treatments.
- Scribble and Glitter, the decorative ones.
The selection is stored as the point you tapped, not as an object number. That distinction sounds pedantic and is not: the segmenter's object ordering is not stable from frame to frame, so a stored index can silently start pointing at a different object partway through a clip. A stored point does not drift like that. The mask is re-derived from the live frame on a cadence and re-matched to that point, so it follows the object as it turns, moves a little and changes shape.
The honest limit
Because the pick point stays where you tapped, this holds an object that stays roughly in place. It is not a motion tracker.
If your subject walks across the frame, the point they used to sit on is no longer on them, and you want the other tool: a label that follows a moving object, which tracks the subject's path and bakes it. Choose by whether the thing moves through the frame or moves in place.
Two smaller things worth knowing. Segmentation only finds objects it considers salient foreground, so a flat sign on a distant wall may not come back as its own object at all. And if you tap near a thing rather than on it, the editor falls back to the largest object it found, which is usually right and occasionally not. Tap the middle of what you want.
Blur is not redaction
This matters more than the technique. A blur is a visual effect on the pixels of the exported file, and heavy blurs on text can sometimes be reversed by someone determined enough. For a password, an address, an account number, blur is the wrong tool. Cover it with something opaque, or reframe the shot so it was never in it.
The same logic applies to sound. Hiding something on screen does nothing about someone reading it aloud in the audio.
For screen recordings specifically, where the sensitive thing is usually text in a window, blurring sensitive information in a screen recording covers a different and generally safer set of moves.
Frequently asked questions
Can I blur one person in a video without blurring the background? Yes, if the editor can select that person as their own object. Tap-to-select does exactly this, and the blur is confined to their shape.
Does this work on older iPhones? The object segmentation this relies on is an iOS 17 feature. On older systems the tap has nothing to pick from.
Does the blur follow the object if it moves? It re-finds the object under the point you tapped every few frames, so it holds through turning and reshaping. It does not follow an object that travels across the frame, which is what tracked overlays are for.
Is blurring enough to hide a password or an address? No. Treat blur as a visual choice, not as redaction. Use an opaque cover, or do not record it.
Can I blur the object and still see the rest of the video normally? Yes. Blur object changes only the pixels inside the object's shape. Spotlight is the inverse if you want the emphasis instead.
Does any of this upload my video? No. The segmentation runs on the phone through Vision, and Zella's editing and export are on device.
Make your next video with Zella.
Record, edit and ship on Mac and iPhone - local, private, free to start.
RELATED