Guide

Blur vs pixelation vs redaction: what actually removes sensitive information?

Use opaque pixel replacement when a screenshot must not retain private text. Blur softens the image and pixelation reduces detail, but both can leave useful clues. Choose the method by what must be removed, then verify the exported file.

Choose by the information you need to remove

Scroll sideways to read all columns.

Method What it doesAppropriate expectation

Blur Mixes nearby pixelsA visual effect, not a guarantee that text is concealed

Pixelation Replaces blocks with a lower-detail patternCan retain structure that distinguishes possible text

Opaque redaction Replaces the entire covered region with a solid valueRemoves those original pixels from a newly flattened image

Redact a screenshot

Compare three saved results from one invented input

Download the same 1200 × 760 fictional dashboard and the three actual public-tool exports. We drew identical regions around the invented name and email and the complete fake key, then exported with Blur, Pixel and Black selected in turn. The linked receipt gives the exact rectangle coordinates, browser, source hash and output hashes.

Blur and Pixel change the covered pixels but can retain visual clues. Black replaces the covered pixels in the downloaded PNG. Reopen the file for the mode you chose and inspect all private locations before sharing. The separate linked experiment compares four invented candidates with code-based blur and pixelation settings; those illustrations are not public-tool exports.

Example billing dashboard with an invented email address, a fake API key and a disabled Export button — Reproducible example input. The name, email, invoices and key are invented for this demonstration.

Fictional dashboard with name, email and fake key processed in Blur mode — Actual public redactor Blur PNG from the same input and two regions.

Fictional dashboard with the same regions processed in Pixel mode — Actual public redactor Pixel PNG from the same input and two regions.

Fictional dashboard with the same regions covered in Black mode — Actual public redactor Black PNG from the same input and two regions.

Download the actual Blur PNG

Download the actual Pixel PNG

Download the actual Black PNG

Check tool settings and file hashes

Download the invented source PNG

Why recognizable structure matters

A short code, repeated label or known set of names gives an observer fewer possible answers. Even a result that looks illegible at first glance can retain differences between candidates.

Our separate experiment compares four invented candidates with known rendering settings. It illustrates retained information without claiming that arbitrary text can always be recovered.

See the experiment and its limits

Flattening matters as much as the rectangle

A black shape in an editable project can be moved or removed. Export a new image in which the pixels have actually been replaced, then inspect that file.

An image-only PDF made from that checked export is a different artifact from printing the original webpage or sharing the editable document.

Create an opaque flattened PNG

Understand the PDF distinction

Finish with a whole-image review

Cover complete values with a small margin and look for repeated occurrences, line wraps and identifying context. Check the exported filename and any separately copied text.

A verified redacted file cannot revoke an exposed real credential. If a key was disclosed, follow the relevant service’s rotation process.

Review API keys and tokens

Use the handoff checklist