You are about to send a screenshot of an invoice, a dashboard or an error message, and part of it should not travel with it: a customer's name, an e-mail address, an account number, an API token. Nearly every screenshot tool offers a blur or pixelate brush for exactly this. The step people skip is asking whether the thing they just did is reversible.
The two failures that actually happen
1. The cover-up that never becomes part of the image
You drag a black rectangle over the account number. On screen it is gone. But in most editors that rectangle is an object floating above the original pixels, and the pixels underneath are untouched. Export a flat PNG and you are fine. Send the editable project file, paste into a document that preserves layers, or hand over the app's own save format, and the recipient can drag your rectangle aside.
This is the same failure that produces the periodic news story about a redacted PDF whose black bars can be selected and copied.
2. The pixelation that can be read back
This one is less well known and more uncomfortable. Standard pixelation takes each block of pixels and replaces it with the average colour of that block. Averaging is a lossy operation, but it is also a deterministic one — and that is the weakness.
If an attacker knows roughly what font and size the text was, they can render a candidate string, pixelate it with the same block size, and compare. Repeat over the plausible strings and the original text falls out. This is not theoretical: published open-source tools do exactly this, and security researchers have demonstrated recovering pixelated text from real screenshots with high success rates on common fonts.
Short, structured values are the easiest of all to recover, because there are so few of them to try: a six-digit code, a date, a four-digit card suffix, a licence plate.
The practical rule: never pixelate text. Pixelation is fine for a face, a photo or a general "this area is not the point" smudge. For anything that is characters, use a solid opaque fill.
The method that holds up
- Capture the screen. Before you do anything else, remember that many tools also save an unredacted original into a history or auto-save folder. Redacting the copy you send does not remove the copy on your disk.
- Cover every sensitive region with a solid, opaque fill. Not a semi-transparent shape, not a light blur, not a mosaic. For text, solid.
- Export a flattened image — PNG or JPEG. This is the step that converts "a shape on top of the text" into "there is no text".
- Reopen the exported file and check it. Zoom into each redacted area. Then look at the rest of the picture, because the leak is usually somewhere you were not thinking about.
The places a screenshot leaks that you did not intend to include: browser tabs with customer names in their titles, an autocomplete dropdown left open, a notification toast in the corner, the taskbar, the window title bar, a second monitor caught in a full-screen capture, and the file name you save it under.
Doing it in Screen Slinger
Screen Slinger is our own product, so treat this section as what it is.
Drag the hide tool over a region and pick one of three styles: solid fill, smooth blur or pixelated mosaic. All three are baked into the image itself rather than drawn on top of it, so what leaves the app in the exported file, on the clipboard and on screen is the same picture — there is no hidden original underneath the shape.
One caveat that applies to us as much as to anyone: the pixelated style averages blocks, like every other implementation, so it is open to the reconstruction attack described above. Use the solid style for text. It fills the region with a single flat colour, and a single flat colour has nothing left in it to reconstruct.
The reason all three styles are baked in matters for the first failure above: annotations in Screen Slinger stay editable objects, but a hidden region is not one of them. It is committed to the pixels.
If you are not using Screen Slinger
The method above is tool-independent, and most tools can do it:
- Windows Snipping Tool — has a built-in redaction option in recent versions. Save the result as a new PNG rather than overwriting.
- Paint — select the region, then fill it with the paint bucket in a solid colour. Save as PNG.
- Any editor with layers — draw the shape, then flatten or "export as" before sharing. The flatten is the whole point.
Whatever you use, the last step never changes: open the file you are about to send, and look at it. Not the editor window. The file.
Written by the team behind Screen Slinger, a Windows screenshot and annotation tool. The advice above applies whichever tool you use, including the ones that compete with ours — the honest comparison is here, and the companion guide on numbered-step screenshots covers the other half of the job. There is also a guide on capturing a long scrolling page.