Screenshot redactor

The one thing to understand before you trust any of this

Redaction fails in a specific, well-documented way: covering text with a black rectangle in a document viewer hides it visually and leaves it present in the file. Anyone can select the rectangle, move it, and read what is underneath. The same trap exists in image editing if you put the mask on a layer and export something that keeps layers.

This tool avoids that class of failure by construction. The pixelation is applied to the pixels themselves and the result is flattened into a single-layer PNG. There is no layer to lift, no hidden text underneath, and no original embedded in the output. Once you download the file, the covered information is gone from it.

One caveat that is yours to handle, not the tool’s: the file you started with still exists on your device with the sensitive content intact. Redacting a screenshot does not delete the original, and if you uploaded the original somewhere before covering it, no amount of blurring afterwards undoes that.

Pixelate or blur — which is actually safer

Pixelate is the correct default. It collapses a region into a small number of blocks, and once flattened the original detail is genuinely not in the file. At high strength it is also instantly legible as “this was hidden on purpose,” which is a useful signal in its own right.

Blur is prettier and weaker. A soft blur at low strength can be sharpened back into legibility by off-the-shelf tools; a heavy blur cannot. If you use blur, use enough of it that the text is unrecoverable rather than merely inconvenient — which is more than most people think they need.

The strength slider is shared between them on purpose. If you want to know whether you used enough: zoom the exported file to 100 per cent and try to read the covered area. If you can guess at it, increase the strength.

What people actually cover, in order

• Names, avatars and @handles in screenshots of other people
• Phone numbers, email addresses and booking references
• Amounts, account numbers and card digits on invoices and receipts
• Room numbers, addresses and licence plates in travel photos
• Anything in a background you did not notice until the screenshot was taken

That last one is the reason the tool exists. Most accidental disclosures are not the thing you were screenshotting — they are the notification banner, the browser bookmark bar, or the name of the person sitting next to you in the frame.

Workflow

Drop the image, drag a box over each thing that must go, and check the count in the status line before you export. Undo removes the last region; clear starts over. Because each region is stored as a rectangle and re-applied on every render, you can change the effect or the strength after drawing them — adjust the slider and the whole set updates.

Nothing is uploaded. The file is read locally, drawn to a canvas in the same page, and exported as a new PNG from that canvas. There is no request in this page that could carry your image anywhere.

FAQ

Is the blurred area recoverable?

At high strength, no — the pixels are averaged and the original detail is not in the output file. At low strength, treat it as recoverable by someone motivated enough. Pixelate is the safer default for anything that matters.

Can I cover a face?

Yes, though a rounded face wants a generous box and high strength. There is no automatic face detection — deliberately, because a detector that misses silently is worse than a box you draw yourself.

Does the export keep the original size?

It keeps the image’s dimensions, capped at 1200 pixels wide so a 12-megapixel phone photo does not become a 6-megabyte PNG. For anything that needs full resolution, do it in a desktop editor.

Where is the file stored?

Nowhere. It stays in your browser tab until you close it, and closing the tab forgets everything.