Guide · Images

Blur, pixelate or black out: what really hides it

By Suprabhat Updated October 11, 2026 How we write these

3 min read

The short answer

Only a solid, fully opaque box saved into the picture removes what is underneath. Blur and pixelation keep a trace of the original, and researchers have recovered pixelated text and matched blurred faces. Use blur or pixelation for faces you just want to soften, and black out anything that must not be read: numbers, passwords, addresses.

What each method leaves behind

Blur averages each pixel with its neighbors. Large shapes survive, which is why a blurred face can still be recognized by people who know it, and why light blur on big text can often still be read.

Pixelation replaces each block with its average color. Every block still depends on what was there, and for text in a known font that is enough: the Depix tool (2020) and Bishop Fox’s Unredacter (2022) both recovered pixelated text from screenshots by trying characters until the blocks matched.

A solid box replaces the pixels with one color. Nothing about the original is left in those pixels to work from.

Faces are different from text

A 2016 study by McPherson, Shokri and Shmatikov showed that machine-learning models could still identify many people from pixelated and blurred faces. For a passer-by in a holiday photo, a strong blur is a reasonable courtesy. For someone whose identity genuinely has to be protected, cover the face completely, and remember hair, clothes, tattoos and the setting can identify someone too.

Mistakes that undo the hiding

A semi-transparent highlighter or marker. Many phone markup tools draw at less than full opacity, and raising the contrast of the picture can bring the text back. Use a fully opaque shape.

Edits that are not saved into the pixels. A box drawn as a separate layer or annotation can be removed by anyone who opens the file in the right app. In PDFs this is the classic failure: a black rectangle over text that can still be selected and copied.

Cropping tools that kept the old data. In 2023 the “aCropalypse” bug showed that Google Pixel’s Markup tool and the Windows Snipping Tool could leave the cropped-away part of a screenshot inside the saved file. Re-saving through a different tool removes such leftovers.

The metadata. A photo of a blurred street sign that still carries the GPS position of the house hides nothing. See what a file carries with the photo metadata viewer.

Doing it properly

In Blur an Image, mark each area and choose Black out for anything that must not be read; the areas are painted into the pixels, the image is saved again and its metadata is removed. For screenshots, Redact a Screenshot finds emails, card numbers and keys for you. To strip location and camera details from photos you are not editing, use Remove Metadata.

Questions

Can pixelated text be unpixelated?

Sometimes. When the font and the background are known, tools have recovered pixelated text by trying likely characters until the pixelation matched. A solid black box cannot be reversed this way.

Is blurring a face enough?

For a stranger in the background of a holiday photo, a strong blur is usually enough. For anyone who must not be identified, cover the face completely and check the rest of the picture for clues.

Does a black box in a PDF hide the text?

Only if the text underneath is removed. Many PDF editors just draw a box on top; proper redaction deletes the characters.