Redacting an image means replacing the sensitive pixels with an opaque block and saving a new, flattened file, so nothing underneath can be recovered. This tool exports a freshly encoded copy with no EXIF, GPS or other hidden fields, and it can spot card numbers, account numbers, contact details and 2-D barcodes on your device to help you cover them.
When you need to redact a picture
Plenty of everyday tasks ask for a photo of a document:
- a landlord, employer or marketplace wants "a picture of your ID" to verify your name, but not your license number, address or date of birth;
- a support agent needs to see a utility bill to confirm an address, not the account number;
- you sell a used item and want to show the receipt without your card digits or loyalty number;
- you post a boarding pass or a concert ticket the night before the trip or the show;
- a form, letter or contract needs a second opinion from a friend, minus the reference numbers.
In each case the goal is the same: share the part that answers the question and remove the rest for good.
What "redacted" should mean
A redaction is only real if the hidden information is gone from the file, not just covered on screen. Drawing a black rectangle in an annotation layer, highlighting text with a marker that is partly transparent, or placing a shape over a PDF without flattening it can all leave the original readable to anyone who removes the layer, raises the contrast or copies the text.
This tool works on pixels, and the cover is permanent in the exported copy. When you export, the image and your boxes are painted together onto a blank canvas and encoded as a brand-new picture. There is no separate layer to lift off, and the covered pixels are not in the file at all.
Steps
- Photograph or scan the document flat and well lit, then open the image here with the file button or by dropping it on the page.
- Press Find text to let the page read the document on your device. It proposes boxes for email addresses, phone numbers, card numbers that pass the Luhn check, IBANs, other numbers of eight or more digits, and 2-D barcodes (QR, Data Matrix, Aztec and the stacked PDF417 type). The reader is downloaded the first time, and its size is shown beside the button.
- Add boxes by dragging over everything else you want hidden: names, ID numbers with letters in them, addresses, signatures, dates of birth, ordinary striped barcodes, photos of faces.
- Leave Cover with on Solid block. Do not use Pixelate or Blur on text or numbers; both are decorative and can be reversed.
- Press Export, open the new copy at full size to double-check every field, then save or share it.
What to cover on common documents
| Document | Usually safe to show | Usually worth covering |
|---|---|---|
| Driver's license or ID card | Your name, the issuing state or country, the photo if identity is the point | License or ID number, date of birth, street address, the barcode or machine-readable zone on the back, signature |
| Utility or phone bill | Your name, the service address if it is being verified | Account number, PIN or security words, usage history, payment details |
| Store receipt | Store name, item, date, total | Card digits beyond the last four, loyalty or member number, any barcode or QR code that looks up the order |
| Shipping label | Carrier and destination city | Full name and street address, phone number, tracking barcode |
| Boarding pass | Airline and flight date | The barcode, booking reference, frequent flyer number, full name |
| Bank or tax letter | Sender and subject | Account and reference numbers, balances, signature |
Boarding passes deserve a special note. The barcode is not decoration: a 2015 investigation by security reporter Brian Krebs showed it can encode the passenger's name, booking reference and frequent flyer number, and that this was enough to open the booking and change or cancel the flights. Cover the whole barcode, not just the printed reference beside it.
The stacked PDF417 barcodes used on boarding passes and on the back of many US licenses are among the codes Find text looks for, along with QR, Data Matrix and Aztec. Ordinary one-line striped barcodes, like the tracking code on a parcel, are not detected; box those yourself.
Make the redaction hold up
- Cover generously. Leave a margin around each field so no stroke of a letter peeks out at the edge.
- Hide the whole item. Hiding half a number or one line of an address can let someone work out what is missing.
- Watch for duplicates. Many documents print the same number twice (a header and a footer, front and back), and IDs repeat details in the barcode.
- Mind the length. A box that hugs a name tells the reader how long the name is. A uniform, wider box gives away less.
- Check the background. Desks, other papers and screens behind the document can show more than the document itself.
What the exported file holds
The new image contains your document with the solid covers burned in, at the original resolution. It does not contain the photo's hidden data: no EXIF camera details or capture date, no GPS position, no XMP or IPTC captions, no maker notes, no embedded preview thumbnail, no Content Credentials. That matters for documents photographed at home, since a phone photo often records exactly where it was taken.
Large scans are previewed smaller, so that the longer dimension stays near 2,048 pixels and the editor responds quickly; your covers land on the full-resolution document at export. If a phone browser cannot encode a scan that large, it saves one at 4,096 pixels on the long side and says so.
Formats and limits
- Images: JPEG, PNG and WebP in all current browsers; HEIC in Safari, which can display it. Other browsers ask you to convert HEIC to JPG first.
- Text finding: the reader uses English language data. Email addresses and the numbers it looks for are made of Latin letters and digits, so it can often spot them on documents written in other languages too. Handwriting, low light, glare, curved pages and decorative fonts reduce accuracy. Always check by eye.
- PDF files: this tool opens images, not PDFs. To redact a PDF page, take a screenshot of it or export the page as an image, redact that, and share the image. If you only need to clear a PDF's author and other document properties, use a separate tool: PDF files: edit or clear PDF properties.
Your document stays on your device
Choosing a file here does not upload it. The page reads the image inside your browser, the text reader runs inside your browser, and the redacted copy is created in memory and handed to you as a download. The reader's program files and English language data come from this site when you first press Find text; your document is never part of those requests. Closing the tab drops everything the page created.
How it is built
Text finding combines the Tesseract OCR engine (compiled to WebAssembly) with pattern rules: email address shapes, phone number lengths and separators, the Luhn checksum for card numbers, the check digits built into IBANs, and any run of eight or more digits. 2-D barcodes are located with the ZXing barcode reader. All of it runs locally. The export re-encodes the image, which is what removes the metadata. You can verify that nothing is sent while you work by following the step-by-step network check.
FAQ
Can a black box in an image be removed later?
Not from the exported copy. The cover replaces the pixels and the file is encoded fresh, so there is no hidden layer and no original underneath. That is different from annotation boxes in PDFs or editors that save layers.
Is it safe to blur a document instead?
No. Pixelated text, and in some cases blurred text, has been reconstructed by public tools when the font can be guessed. Use Solid block for every field on a document.
Does Find text send my document to an OCR service?
No. The OCR engine and its language data are downloaded to your browser and run there. The document itself is not transmitted.
Will it find every sensitive number?
Its rules cover addresses in email form, phone numbers, Luhn-valid card numbers, IBANs, digit runs of eight or more, and 2-D barcodes. Names, short reference numbers, ID numbers that mix letters and digits, and striped barcodes need boxes drawn by hand, and you should check the result by eye in every case.
Does the redacted image still show where the photo was taken?
No. Coordinates and the rest of the EXIF block stay behind in the original; only what is visible in the picture remains.
Can I redact a PDF here?
Not directly. Save the page as an image, redact the image, and share that. For a PDF's document properties, a dedicated PDF tool is linked above.
Sources
Checked 2026-10-11.
- Krebs on Security, "What's in a Boarding Pass Barcode? A Lot" (October 2015)
- Tesseract OCR documentation: the open-source engine used for on-device text finding.
- ZXing-C++ barcode library: the reader used to locate QR, Data Matrix, Aztec and PDF417 codes.
Last updated: