How to Blur Part of a Screenshot (and Why That's the Wrong Tool)
Blurring or pixelating private details in a screenshot before sharing it can be reversed. Here is why, and the one technique that actually removes the information for good.
If you searched for how to blur part of a screenshot, you probably want to hide a name, an email, an account number, or something else private before sharing the image. Blur is the wrong tool for that. It looks like it removes the information; it does not reliably do so.
Why blur and pixelation are not “gone,” just obscured
Blurring averages nearby pixels together. Pixelating replaces a block of pixels with a single averaged colour. Both are lossy — you cannot mathematically recover the exact original from the result — but lossy compression and secure erasure are not the same thing. The blurred or pixelated region still carries a statistical fingerprint of what was underneath it, and how strong that fingerprint is depends entirely on the blur radius relative to the size of the content.
Small text is the worst case. A password field, a card number, or a name rendered at a normal reading size has sharp, high-contrast edges that a moderate blur does not fully destroy — it smears them, but the underlying structure (letter spacing, stroke width, character boundaries) survives well enough that reconstruction is a known, repeatedly demonstrated attack, not a theoretical one. This is not a new finding — it is well-documented enough that it should be treated as a settled fact, not a maybe: treat any blurred or pixelated text in a screenshot as recoverable, not removed.
The fix: cover it, don’t blur it
There is a version of this that actually works, and it is simpler than blurring: draw a fully opaque box over the sensitive area instead of a filter over it.
The reason this is safe where blur is not: opacity is binary. A pixel is either covered by the solid fill or it is not — there is no partial, statistically-recoverable middle ground the way there is with a blur radius. Security researchers recommend exactly this for redacting documents and screenshots, and it is the only approach on this list that removes the original information from the exported file rather than degrading it.
Doing it
- Load the screenshot into an editor that draws directly onto the image rather than only applying filters.
- Select a filled rectangle tool — not an outline, a solid fill.
- Draw it over the sensitive area with a small margin on every side, so the box extends slightly past the text rather than stopping exactly at its edges.
- Zoom in and check the boundary. A sliver of visible text at the edge of the box is the most common way this fails.
- Check the rest of the image for the same information appearing again — a URL bar, a notification, a sidebar — before you export.
- Export. The covered region is gone from the file, not just visually hidden in your current view.
ShotEdit’s screenshot editor has a filled rectangle tool built for exactly this, works entirely in your browser with no upload, and exports without a watermark — which matters here specifically, because the screenshot you are redacting is by definition one you did not want to expose in the first place, and running it through a server before you have removed the sensitive part defeats the purpose.
If the information doesn’t need to be in the shot at all, crop it instead
Covering is for information that has to stay visible in context — a UI element you are demonstrating, with one field redacted. If the sensitive detail is incidental — a name in a browser tab, a notification banner that happened to pop up during capture — cropping it out entirely is simpler and just as effective, since a cropped region is not present in the exported file at all.
Decision table
| What you’re hiding | Right technique | Why |
|---|---|---|
| A field inside a UI you’re demonstrating | Filled rectangle over just that field | Keeps the surrounding context visible |
| Something incidental to what you’re showing | Crop it out of frame | Simpler, and nothing to double-check |
| Short, structured data (password, card number, ID) | Filled rectangle only, never blur | Highest reconstruction risk if blurred |
| Anything at all | Not blur or pixelation | Both are demonstrably reversible |
The one habit worth keeping from all of this: if you catch yourself reaching for a blur tool to hide something private, that is the signal to switch to a solid box instead — the intent is the same, only the box actually finishes the job.
Frequently asked questions
Is blurring a name or email in a screenshot before sharing it safe?
Not reliably. Blur and pixelation are lossy, but lossy is not the same as gone — the underlying pixels still carry a statistical relationship to the original text, and that residual structure is exactly what makes reconstruction attacks possible when the blur radius is small relative to the character size. A light or moderate blur over small text is the riskiest case.
Can a pixelated screenshot actually be un-pixelated?
Security researchers have repeatedly demonstrated reconstructing text from pixelated or lightly blurred screenshots, particularly for short, structured content like passwords, card numbers, or code — and there have been real, publicised cases of redacted documents and screenshots being reversed this way. The technical detail is not the point here; the fact that it is a demonstrated, repeated category of failure is.
If not blur, what should I use to hide information in a screenshot?
A fully opaque, solid-colour box drawn directly over the information. There is no partial version of an opaque fill — either a pixel is covered or it is not — so there is no residual signal for anything to reconstruct. This is the standard recommendation from security researchers and the only method that removes the underlying data rather than obscuring it.
Does covering information with a solid box leave any trace of the original text?
No, as long as the box is genuinely opaque and fully covers the sensitive area with a small margin. Check that the box does not end exactly at the edge of the text — a pixel or two of visible antialiasing at the boundary is a common way partial coverage slips through unnoticed.
Is it enough to redact one instance of the sensitive text if it appears twice in the same screenshot?
No. Scan the entire image before exporting — the same email address, name, or ID number showing up unredacted in a sidebar, a URL bar, or a notification elsewhere in the same screenshot is one of the most common ways redaction fails, and it happens because people fix the obvious instance and stop looking.