There are five ways to do this and they differ in one respect that matters more than ease of use: whether your video is sent to somebody else to be processed. Here is what each method costs you, and how to pick before you hand over a file you cannot take back.
The four questions that decide it
Every comparison of these tools eventually turns into a list of menu paths, which go out of date within a year. These four do not.
- Does it upload?
- Most free online blurring tools send your video to a server and process it there. That is a copy of the unredacted file, on somebody else’s disk, for a retention period described in a policy you did not read. For an ordinary clip this is fine. For anything sensitive it is the whole problem.
- Does it track?
- A cover that does not follow the face has to be keyframed by hand. Heads move constantly, and the frames where a manual mask drifts are exactly the frames somebody screenshots.
- Can it be undone?
- A blur can often be partly reversed, and any obfuscation can be matched against a short list of suspects. The maths, and what survives each treatment.
- What does it cost you?
- Time, quality, or a watermark. A tool that is free, fast and unrestricted is usually charging in a currency that is not money.
The methods, in order of who they suit
1. A phone editor with a mosaic effect
Mobile editors will drop a mosaic onto a region and track it through the clip, and they do it on the phone. This is the quickest route for a short casual clip, and the tracking is usually adequate for a face that stays roughly front-on.
Two things to watch. The tracker gives up on turns and fast movement, and it does not tell you when it did — you find out by scrubbing the export frame by frame, which almost nobody does. And the free tiers of these apps often export at reduced resolution or add a watermark, so check the file you got rather than the preview you watched.
2. A desktop editor you already know
Premiere Pro, Final Cut and DaVinci Resolve all do this properly: a mask on the face, a planar or point tracker to carry it, a mosaic or blur inside the mask. Resolve is free and local, and if it is already installed it is a genuinely good answer — better than most of the web tools, because nothing leaves the machine and you can see exactly what the mask is doing on every frame.
The cost is time and attention. Every face is a separate mask, every tracker needs correcting where it slips, and a ten-minute clip with three people in it is an afternoon. If you are doing this once, that is fine. If you are doing it weekly, it is not.
3. An online blurring tool
Upload, wait, download. These are the tools that rank first for this search, they are easy, and for a video of your dog they are the sensible choice.
They are the wrong choice for anything where being identified has a consequence, and the reason is not that the companies are untrustworthy. It is that you have created a second copy of the unredacted video, on infrastructure you do not control, and no amount of deleting it afterwards changes the fact that it existed there.
4. A browser tool that runs locally
Recent browsers can decode, process and re-encode video on the device using the same hardware the camera app uses. Nothing has to be uploaded to do any of it, and nothing has to be installed either. This is what Sitr is: the file is opened, every face is found and covered, and the result is saved — all inside the tab.
The claim is checkable, which is the only reason to make it. Turn off your network connection and run a file through. It finishes exactly the same, because there was never anything to send.
5. Doing it after upload, on the platform
Some platforms offer to blur faces in a video you have already uploaded. Whatever else is true of those tools, they cannot solve this problem: by the time the option exists, the original is already on their servers.
Redact before you upload. Everything else is redacting the copy.
If you are doing it by hand, do these four things
- 01
Cover more than the eyes
An eye bar defeats a stranger and a face-matching system. It does not defeat somebody who knows the person, because a jaw, a hairline and a mouth are plenty. Decide who you are hiding from before you choose how much to cover.
- 02
Use a mosaic or a solid shape, never a blur
A blur keeps the face in the file in smeared form. A solid fill keeps nothing. If the tool offers both, there is no situation where the blur is the better choice.
- 03
Step through the export, not the preview
The preview is the tool telling you what it intended. The export is what people will download. Scrub the finished file frame by frame through the moments where the head turns, because those are where a tracker lets go.
- 04
Strip the metadata
A video file carries creation times and, from many phones, the location it was recorded at. Covering the face and leaving the GPS coordinates in the container is a common and complete failure.
Questions people ask next
How do I blur a face in a video on an iPhone?
There is no built-in way. The Photos app can mark up a still image, but it will not blur or cover a face in a video, and iMovie has no face tracking — you would be masking each moment by hand. The practical options on a phone are a video editor with a mosaic effect and a tracker, or a browser tool that does the tracking for you. A browser tool has the advantage that nothing has to be installed and, if it runs locally, nothing has to be uploaded.
Can I blur a face in a video for free?
Yes, several ways. DaVinci Resolve is free and does it properly, at the cost of a real learning curve. Mobile editors offer a free mosaic effect with tracking. Sitr covers your first files free and runs in the browser. The thing to check in every case is not the price but whether the video is uploaded to somebody else to do the work.
Does the blur have to be re-applied after uploading to TikTok or Instagram?
No — and it must not be. Cover the faces before you upload. Every platform re-encodes what you send it, so the file that ends up on their servers is derived from the one you handed over, and the one you handed over is the one that existed unredacted. Redaction applied after upload protects the copy that is public, not the copy they already have.
Will blurring a face reduce the video quality?
The covered area is destroyed on purpose, which is the point. What people usually mean is whether the rest of the video is degraded, and that depends on the export: any tool that decodes and re-encodes a video loses a little, so the thing to check is that it exports at the original resolution and a sensible bitrate rather than downscaling to 720p and stamping a watermark on it.
What is the difference between blurring and pixelating a face?
A blur averages each pixel with its neighbours and keeps all the information in a smeared form, so it can be partly reversed. Pixelation averages whole blocks down to one colour, which throws most of it away and cannot be inverted — though the surviving block colours can still be matched against a list of suspects. A solid bar leaves nothing at all.
Or let it do all four for you.
Sitr finds every face, covers it, follows it through movement and turns, then decodes the finished file and checks every frame again — flagging anything it could not clear rather than letting you find out later. In the browser, on your own device.
2 files freeNo cardNothing uploaded