Drag boxes over names, card numbers, faces or number plates, and choose how each is hidden. The saved image contains only the hidden version, and the photo's location and camera details are removed. Everything stays on your browser.
Blurring and pixelating keep some information from the original. For short, predictable text like a card or phone number, someone can try every possibility until the blur matches. Here's the same card number hidden three ways at the strength you chose.
Phone photos carry hidden details written by the camera. Covering a street sign achieves little if the file still records the exact GPS position. Saving from this tool writes a fresh image with none of it.
An attacker doesn’t need to “unblur” anything. If they know the font and roughly where the text sits, easy for a screenshot, they render every possible value, hide each one the same way, and keep the one that matches best. This runs that attack for real in your browser.
A cover drawn at less than full opacity, or a dark marker in multiply mode, scales the text down instead of removing it. The difference may be only a few brightness levels: invisible to you, but stretching the covered area to full contrast brings it straight back. Saving as JPEG adds noise, which helps a little but not reliably.
Most redaction failures aren’t clever maths. The original was simply still there: under a layer, under a low-opacity box, in a thumbnail or past the end of the file.
Neighbouring tools, with a typical result for each.