Can the original content be recovered from a redacted area?
No. Redaction here genuinely replaces the pixels in that region with block averages, so the original pixel data simply is not present in the exported image — it is not an opaque sticker laid on top. That distinction matters: some tools only draw a covering layer, or apply a reversible blur that can be partially undone by processing the image. One caveat worth stating plainly: redaction only covers the regions you actually drew over, so give the export one last look before sharing it.
Why does redaction in particular make such a point of not uploading?
Because what you are hiding is usually the very thing that should never leave your machine — a password, an API key, an ID number, a card number, a private conversation. If a redaction tool asks you to upload the un-redacted original to its server and sends back a censored copy, then the sensitive original has already left your device in full. This tool works entirely inside your own browser; not a single byte of the original is transmitted, and you can even go offline and keep using it.
What if I draw something wrong — can I move or delete one annotation?
You can step backwards with Undo, or use Clear all to start the annotations over (your underlying image is untouched). Selecting an individual finished annotation to drag or delete it is not supported yet — for now, undoing and redrawing is the fastest fix for a misplaced mark.
Which formats can I import, and what do I get out?
Anything your browser can decode — JPEG, PNG, WebP, GIF and so on — including screenshots pasted straight from the clipboard. Exports are always PNG at the original resolution with no lossy compression, and nothing is ever stamped on them: no watermark, no branding.