Sitr

Cover every face in a photo, or a whole camera roll.

Drop in one photograph or a whole camera roll. Every face is found and covered on this device, at full resolution, and nothing is uploaded to do it — so the unredacted originals never exist anywhere but your own phone.

2 files freeNo cardNothing uploaded

Blurred

Recoverable

Covered

Nothing left to recover

A roll at a time

Two hundred photographs in one run, one set of settings, and a list at the end of the few worth a second look.

The coordinates go too

The covered image is written fresh, so the location your phone wrote into the original does not travel with it.

Full size, no watermark

The resolution you gave it, on the free files as well. A plan changes how many you may make, not what they look like.

How it goes

  1. 01

    Drop the photographs in

    JPEG, PNG or WebP — and HEIC in Safari, which is the browser that decodes what an iPhone saves. One photograph, or as many as you have. The browser is opening local files; nothing is sent.

  2. 02

    Every face in every image is covered

    Each photograph is decoded, the faces found, the eyes covered, and the result re-encoded — all on your own hardware, one after another.

  3. 03

    The uncertain ones are flagged

    Anything the checking pass could not clear confidently is listed at the end with the image it came from, so you review three rather than all of them.

  4. 04

    Download the batch

    Save them all at once, or send them straight to a linked account. Even then the file goes from your browser to the platform and never through us.

The part people get wrong after the faces

A photograph from a phone usually carries an EXIF block, and on default settings that block contains the exact coordinates the shutter was pressed at, the time to the second, and the device it was taken on. None of it is visible when you look at the picture. All of it is one command away for anybody who downloads it.

A carefully covered face in a file that still says where and when is not a redacted photograph. It is a redacted photograph with an address on it.

Because every image here is written fresh rather than edited in place, the original metadata block does not travel into the file you download. That is a consequence of how the pipeline works rather than a feature bolted on, which is the version of this you want — but it is worth confirming for yourself on the first file, the way you would confirm anything else that matters.

exiftool on the downloaded file will tell you in one line. It is ten seconds, once, and it settles the question for every file after it.

What it takes, and what it gives back

Formats in
JPEG, PNG, WebP everywhere; HEIC in Safari, which is the browser that decodes it.
Resolution out
The one you gave it. No downscale, no watermark, free files included.
Batch size
As many as the device will hold. They run in sequence, so a large batch takes longer rather than failing.
Covers
Frost, bar or streak — all three replace the pixels rather than averaging them. More about the bar.
Where it runs
Chrome, Edge, or Safari 16.4 and later, on a laptop or a phone. Works offline.
What we receive
Nothing. There is no server in the path.

Who this is actually for

  • Anybody posting a group photograph who did not ask the other eleven people in it whether they wanted to be on the internet.
  • Journalists and researchers with a folder of faces that cannot be published and cannot be uploaded to a web tool either.
  • Anyone filing or sharing evidence — a claim, a report, a complaint — where a bystander’s face has no business being in the bundle.
  • Teachers, clinicians and carers, for whom a photograph of the room is a photograph of people who did not consent to it.

Questions

Can I do a whole camera roll at once?

Yes. Drop in as many photographs as you like and they are covered in one run with one set of settings. Anything the checking pass is unsure about is flagged at the end, so you look at three of them rather than all two hundred.

Are the photos uploaded?

No. Every image is decoded, covered and re-encoded inside the browser tab on your own device. There is no upload step and no server in the path, which you can confirm by disconnecting from the network and running a batch — it finishes exactly the same.

Does it handle HEIC from an iPhone?

In Safari, yes, and HEIC is what an iPhone saves by default. Chrome and Firefox will not decode HEIC at all — that is the browser, not the tool — so on those, open the page in Safari instead, or set the camera to Most Compatible so it writes JPEG. JPEG, PNG and WebP work everywhere.

Does it remove the location data?

Yes, as a consequence of how it works rather than as a feature bolted on. The covered photograph is encoded fresh from a canvas, so it inherits nothing from the original container: no EXIF, no GPS coordinates, no camera serial, no timestamp and no embedded thumbnail. That last one has caught out other redaction tools — an EXIF thumbnail is a second, smaller copy of the picture, and more than one tool has shipped an uncovered face inside it. Orientation is the one thing kept, and it is baked into the pixels so the photograph does not come back sideways.

What if a face is turned away, or partly hidden?

A profile, a face in shadow and a face behind a hand are all harder than a front-on portrait, and where the detector is confident about a head but not about the eyes it fails closed: it covers the whole head rather than placing a bar where it is guessing. Anything it is unsure about is shown to you rather than quietly passed.

Can I choose what the cover looks like?

Three styles: a frost, a solid bar, and a streak. All three replace the pixels underneath rather than averaging them, so the choice is about how the picture reads, not about how well it hides anybody.