GUIDE · THE ONE WORTH READING TWICE

How to redact a screenshot properly

Short version: for text, blur and pixelation are decoration. Use bars.

The uncomfortable fact first

Pixelated text can be recovered. This isn't theoretical: publicly available tools reconstruct text from pixelation by brute force — render every candidate string in the right font, pixelate it the same way, compare blocks, repeat. Security researchers have demonstrated it on real "redacted" documents, and moderate blurs fall to similar techniques. If you pixelated an API key at block size 8 and posted it, you should rotate the key.

Pixelation and blur are fine for faces at strong settings — a face isn't drawn from a 90-character alphabet in a known font. Text is, and that's exactly what makes it recoverable.

Rule: if the sensitive thing is text — names, emails, keys, order numbers, balances — redact with the solid BAR. Bars replace the pixels. There is nothing left to recover.

The method

  1. Paste the screenshot into the tool with Ctrl+V — no need to even save it to a file first. Processing is local to your browser tab; the image never touches a server. For sensitive content, that's not a nicety, it's the point.
  2. Bar every piece of sensitive text. Select BAR, drag boxes. Match the bar color to the app background (color picker in the toolbar) if you want it subtle — a dark-mode-colored bar reads cleaner than default black on a dark screenshot. Subtle or not, it's equally unrecoverable.
  3. Sweep the edges for the leaks you didn't screenshot on purpose:
    • Browser tab titles (and the count — "437 unread" tells a story)
    • The address bar: URLs carry account IDs, tokens, internal hostnames
    • Notification toasts that slid in as you took the shot
    • Taskbar / dock / menu bar clock — timestamps have contradicted more than one alibi
    • Usernames in window titles and terminal prompts
    • Autocomplete dropdowns and recently-used lists
  4. Export a fresh PNG. The download is a flat re-encoded image: no layers, no edit history, no EXIF. What you see is the entirety of what you ship.

Why not just draw a rectangle in the doc?

Because in layered formats the rectangle is an object on top of the text, not a replacement for it. PDFs, Word docs, and slide decks with "redactions" that were just black shapes have leaked selectable text repeatedly, including in high-profile court filings — copy, paste, there's your secret. A flattened image with a bar has no such underneath. If you must redact a PDF, use an actual redaction feature that removes content; if it's going out as an image anyway, flatten first and bar it here.

When blur or pixelation is fine

  • Faces — at heavy settings. See the face guide for what "heavy" means.
  • Visual context — a cluttered desktop behind the window that's merely embarrassing, not secret.
  • Aesthetics — softening a background so the shared part pops. Decoration, in other words. Which is what blur is.
Redact it now
Paste with Ctrl+V · bars, blur, pixelation · nothing uploads
OPEN TOOL

Questions

Can pixelated text really be recovered?
Yes — with public tools, given the font and block alignment, which screenshots conveniently standardize. The attack is a search over candidate strings, not magic enhancement. Bars are immune because the information is gone, not scrambled.
I already posted something with pixelated text. Now what?
Treat the text as exposed and act accordingly — rotate the key, change the password, assume the number is known. Then repost with bars if the image still needs to be up. (And per our no-drama policy: we never saw it; nothing you open in BlurThat reaches us.)
Is a colored bar as safe as a black one?
Yes. Any solid fill replaces the underlying pixels completely. Color is purely cosmetic — match the app background if you want the redaction to look tidy.

Related