Blur or pixelate areas

It depends on the mode, and it is worth knowing before you post anything. Pixelate at 100% leaves each area a single flat colour: there is nothing underneath, not even for software. Below that — and with blur at any strength — some of the original shape survives, and for short, predictable text such as a number plate, someone determined enough could get close. If what you are hiding matters, pixelate and push the strength to the top.

No. You mark the areas yourself by dragging on the image. Detecting faces automatically needs a model, which would mean downloading one into your browser or running it on a server — exactly what this tool avoids. Nothing here leaves your machine.

Yes, up to 20. Drag outside the existing ones to add another, inside one to move it, and pull its corners to resize. Each area has its own delete button. The mode and the strength apply to all of them at once.

No. The areas are rewritten inside your browser and the file never leaves your device. You can check it with the Network tab of your developer tools open, or by disconnecting the internet once the page has loaded: the tool keeps working exactly the same.

How much gets covered, measured against the size of the area rather than the size of the photo. That is why the same 60% covers just as much on an 800-pixel image as on a twelve-megapixel one: with pixelate it sets the block size, with blur the radius.

Yes. The areas you mark are applied to every image on the list, in the same place on each. That works well for a series shot from one spot; if the framing changes between photos, check each result before you publish them.

Recreating the image drops most of that data, but this tool is not built for it and does not promise it. If what worries you is where the photo was taken or which camera took it, run the file through "Remove metadata", which is the tool for that job.