Image Blur & Redaction
Blur, pixelate or black out a region. The solid block is the only one that actually hides what was underneath.
Scaled to this image, so it means the same on any file.
For a soft background rather than hiding one detail.
Drop an image here
or paste a screenshot with Ctrl+V. Nothing is uploaded.
Everything here runs on your device. Nothing you enter is uploaded or stored.
There is a safety problem at the centre of every blur tool, and it matters more than any feature: blur and pixelation are not secure redaction. This tool says so on the page, offers a mode that actually works, and makes that mode the default.
How to use it
- Drop an image onto the panel, press Choose an image, or paste a screenshot.
- Leave the mode on Solid block if you are hiding real information. Pick Blur or Pixelate only for cosmetic softening.
- Set the region in pixels, or tick Whole image.
- Press Download.
Why pixelation can be undone
Both blur and pixelation are deterministic functions of the pixels they replace. Pixelating a region averages each block, and that average still correlates with whatever was underneath it. When the possible contents are constrained — the digits of a card number, a postcode, a name from a known list, a licence plate — the attack is straightforward. Render every candidate in the same font at the same size, apply the same pixelation, and compare. The match is usually unambiguous.
This is not theoretical. It has recovered real redacted numbers from published documents repeatedly, and there are tools that automate the whole process against pixelated screenshots. Blur is weaker still in some ways: a Gaussian blur is a linear operation, and linear operations have inverses.
A solid fill has no such weakness, because there is nothing to invert. The pixels are overwritten with a flat colour, and the information is simply gone from the file. That is the difference between hiding something and appearing to.
Strength that means the same thing everywhere
Blur radius and pixel block size are both derived from the shorter edge of the image rather than being absolute pixel values. This is what makes one setting usable across a mixed batch: a fixed 20-pixel radius destroys a small thumbnail and does nothing visible to a phone photo, so the same slider position would produce wildly different results depending on the file.
Pixelation is floored at a two-pixel block, because a one-pixel block is the original image, and a slider whose first position does nothing looks broken.
The block count
Under the pixelate mode the tool reports how many distinct blocks your region has been reduced to. This is deliberately concrete. Telling someone a redaction “may be reversible” is easy to ignore; telling them their redaction preserved 312 averaged values is not. Fewer blocks means less surviving information, but even a few hundred can be enough for short, constrained text.
Noise, and the mode that looks strongest but is not
Scrambling a region with random pixels looks more thoroughly destroyed than a blur, and it is stronger — but the underlying brightness still shows through, because random noise added to a value averages back towards that value. Over a large enough area, the original structure is recoverable. It is labelled cosmetic for the same reason as the other two.
What happens to your file
Everything runs on your own device using Canvas. The image is never uploaded, and the exported file contains only the covered version — there is no hidden original layer inside it, which is a real hazard with some editing formats.
Questions
Can blurred or pixelated text be recovered?
Often, yes. Both effects are lossy but entirely deterministic transforms of the original pixels, so an attacker who knows the character set can render every candidate, apply the same pixelation, and compare against your image. This has been used to recover real card numbers and addresses from published photos, and automated tools to do it exist. For anything that matters, use the solid block.
Why does the tool default to a solid black box instead of a blur?
Because the safe option should not have to be discovered. Most people reaching for a blur tool are trying to hide something rather than make a background pretty, and a tool that lets someone publish a blurred passport number without a word is doing them harm. Blur and pixelate are still here, labelled cosmetic.
What does the block count under the pixelate mode mean?
It is how many distinct averaged squares survive inside your region, which is the actual measure of how much information is left. Four blocks over a six-digit number leaves an attacker with about a million candidates to test, which is instant. A few hundred blocks is still often enough to reconstruct short text, which is why the number is shown rather than described.
Does the strength slider mean the same thing on every image?
Yes, because it is scaled to the shorter edge of the picture rather than being a fixed pixel radius. A 20-pixel blur obliterates a 400px thumbnail and is invisible on a 4000px photo, so a fixed radius would make the same slider position mean something different for every file you open.
Last updated