Blur / Pixelate (redaction)
Redact part of an image — and see why blurring is not redaction.
How this processes your file
You drag over the areas to cover and each is applied to a canvas, which is then re-encoded as a new image. Two things about that matter more than the drawing. The first is the method: a solid fill replaces the pixels, while blurring and pixelation rearrange them — and rearranged information has been recovered in practice, by generating candidate text, pixelating it the same way, and matching. The second is the compositing. Redaction fails most often not because the covering was weak but because the original survived underneath as a separate object, which is what happens in a PDF or an SVG. Here the pixels are overwritten before export, so the downloaded file is all there is.
What the results mean
- Solid fill
- Replaces the pixels. The only method here that removes information.
- Pixelate
- Averages blocks. Recoverable for text in a known font by brute-force matching.
- Blur
- Redistributes detail rather than discarding it. Recoverable in some circumstances.
- Pixels overwritten
- Composited and re-encoded, so nothing survives underneath as a separate layer.
Common problems and fixes
- Is blurring enough to hide a face?
- Not reliably. Swirl obfuscation has been reversed to identify someone, and blurring is weaker still. Use a solid fill for anything that would matter if it came back.
- I blacked something out in a PDF and it was still readable
- The rectangle was drawn over the text rather than replacing it, so the text object survived and could be selected through it. That is the commonest redaction failure there is.
- The metadata is still in my redacted file
- Redaction covers pixels, not metadata. Strip that separately with the EXIF remover.
Frequently asked questions
Can blurred or pixelated text be recovered?
Yes, in the right conditions. If the font and layout are known, candidate text can be pixelated the same way and matched against the image until it fits. This has been demonstrated on real redactions.
Why is a solid fill safer?
Because it discards the information rather than transforming it. There is nothing left to reverse — the pixels underneath are gone, not rearranged.
Does it matter what format I export?
It matters enormously in general. A raster export bakes the redaction in; a PDF or SVG can keep the original as a separate object beneath the box. Everything here is composited to pixels before export, so that failure cannot happen.