Drag over the parts of a photo you want hidden and download the result. Three methods, and they are not equally safe — which most tools in this category do not tell you. A solid block replaces the pixels, so there is nothing left to recover. Pixelation and blurring only average them, and averaging can be undone: researchers have recovered pixelated text by rendering candidate strings through the same pixelation and matching, which works because the character set is small and the process is deterministic. For names, account numbers, addresses or faces, use a solid block. The image never leaves your browser.
| Method | Recoverable? | Use for |
|---|---|---|
| Solid block | No | Names, numbers, addresses, faces |
| Pixelate | Yes — demonstrated on text | Visual effect only |
| Blur | Yes — demonstrated on faces | Visual effect only |
No. It is edited entirely in your browser, which matters for exactly the photos people censor.
No, not for anything that matters. Pixelated text has been recovered by published techniques: the attacker renders candidate strings through the same pixelation and compares, which works because there are few possible characters and the process is deterministic. Blurring has similar weaknesses. Use a solid block.
Because it is the right choice when you want the viewer to see that something was there and roughly what kind of thing — a face in a crowd shot, a background screen. It is a visual effect, not a security measure, and the tool says so.
Yes. Regions are stored as proportions of the image, so a mark drawn on the preview applies precisely to the full-size export rather than at preview scale.
With a solid block, no — those pixels are overwritten before encoding. Be aware the original file on your device is unchanged; this produces a new one.
This tool re-encodes the image, which drops most metadata as a side effect. If removing metadata is the goal, the Image Metadata Remover does it losslessly.
Three methods are offered and they are not equally safe, which the tool states rather than leaving to assumption. A solid block overwrites the pixels: there is nothing underneath and nothing to recover. Pixelation and blurring only average the pixels, and averaging is reversible in the cases that matter most. Published work has recovered pixelated text by rendering every candidate string through the same pixelation and matching the result — the character set is small and the process is deterministic, so this is a search rather than a reconstruction. Faces and number plates have been recovered from blurs by similar means. That is why solid is the default and why the warning appears the moment another method is chosen. Pixelation is implemented by scaling the region down to a small canvas and back up with image smoothing disabled, which genuinely quantises the detail — drawing large squares over the top would merely hide it while leaving the original pixels in the file. Regions are stored as percentages, so a mark placed on a preview a few hundred pixels wide applies accurately to a full-resolution export. Everything happens in your browser, which is the point for photos containing documents or faces.