How to add a watermark to a PDF or a photo

A watermark is an extra layer drawn over the content that is already there. On a PDF, the text is added as real vector type on top of each page, with the opacity you choose — so it stays sharp at any zoom, adds a few kilobytes at most, and leaves the original page objects untouched. On a photo there is no layer to add to: the text is painted into the pixels, which makes it part of the image from the moment you save it.

The quickest route here: for a document, open Add Watermark to PDF, type the Watermark text, pick Once, centred or Repeated across the page, then set the font size, angle, colour and opacity and click Add watermark. For pictures, Add Watermark to Image takes one photo or a batch, with five fixed positions plus Repeated diagonally. Both run in your browser; the files are not uploaded.

The three ways to watermark, and when each one makes sense

Method	Good for	Watch out for
A browser tool like this one	A quick text stamp on every page, or a batch of photos, on any device	Text only — neither tool places a logo or an image
Desktop software (Acrobat Pro, LibreOffice, ImageMagick, GIMP)	Image watermarks, per-page control, scripted batches	Acrobat Pro is a paid subscription; ImageMagick is a terminal tool
Built-in OS features	—	None of macOS, Windows, iOS or Android has a watermark function

That last row is not a hedge. Preview, Edge, the Files app and Google Drive can all view and annotate, but none of them will stamp text across every page for you. The nearest built-in approach is to add a header or a background in the application that produced the document — a watermark in Word or a diagonal text box in LibreOffice Writer — before exporting to PDF.

Choosing settings that actually work

• A document status (DRAFT, CONFIDENTIAL). One large line across the middle: Once, centred, 45°, light grey, opacity around 20%. The default 48 pt is about 17 mm of type height, ascenders to descenders, which reads at a glance without swallowing the text under it.
• A traceable copy. Put the recipient's name and the date in the text — "Copy for J. Smith, 24 September 2026" — and choose Repeated across the page. If the file turns up somewhere it should not, you know which copy it came from. Repeated rows are spaced at five times the font size, so at 48 pt you get about four bands down an A4 page.
• Photos you publish. A corner mark is discreet and trivially cropped away. Repeated diagonally is much harder to remove, at the cost of being more intrusive — the usual trade-off for photographers' proofs. The image tool sizes the text as a percentage of the shorter side (2–15%, default 5%), so a 4000 × 3000 photo gets about 150 pixels of type and a square 1080 × 1080 post gets about 54 — the mark looks the same across a mixed batch.
• Check it before you send it. Open the result and read a page of body text, or zoom into a photo. A very light setting — the PDF tool goes down to 5% — can vanish on a printout; above 40% the mark starts competing with the content.

What a watermark does — and what it does not

A visible watermark is a signal, not a lock. It states a document's status, reminds a reader that the file is not theirs to reuse, and — when it carries a name or a date — makes a leaked copy traceable back to one recipient. Those are real, useful effects, and for drafts circulating inside a company they are usually enough.

What it is not is protection. Anyone with a PDF editor can delete a text object drawn on top of a page, and anyone with an image editor can clone or crop a mark out of a photo. A repeated pattern takes more effort than a single corner mark, and a mark laid across the subject of a photo takes more effort still, but none of this is a barrier to someone who wants the file. Treat a watermark as a deterrent and a label.

If the content genuinely has to stay private, the tool for that is encryption: Protect PDF locks the file with a password so it cannot be opened at all without it, and how to password-protect a PDF covers how to share that password safely. Watermark first, protect second — an encrypted file cannot be edited, so the watermark tool needs the unprotected version.

Flattening, and the order to do things in

If the PDF is a filled form, flatten it before watermarking. Flattening turns the form's answers into ordinary page content, so they can no longer be edited and they display in every viewer, including the mobile apps that ignore form fields. Watermarking a form does not flatten it: the two are separate operations on separate layers.

There is one way to make a PDF watermark much harder to peel off, and it has a real cost. Compress PDF in strong mode renders each page to a JPEG image — at 150 DPI an A4 page becomes roughly 1240 × 1754 pixels — and the watermark becomes part of that image, with nothing left to select and delete. The same operation destroys the text layer, so the document is no longer searchable, no longer selectable, and no longer readable by a screen reader. Use it for a proof or a visual sample, not for a document anyone needs to work with.

Common problems, and what they mean

Accented characters came out wrong, or as question marks. The PDF watermark uses the standard Helvetica Bold font, which covers Latin characters including accents. Cyrillic, Greek, Arabic and CJK characters are not in it, and unmapped characters are replaced by "?".

The watermark hides the text underneath. Lower the opacity, reduce the font size, or switch from repeated to centred. On a dense page, 20% grey at 45° is a reliable starting point.

Only some pages should be marked. The PDF tool stamps every page. To mark a subset, split the document, watermark the part you need and put it back together.

The mark is invisible on a photo. White text on a bright sky disappears. The image tool adds a soft drop shadow to help, but a dark colour or a corner with less going on will do more.

A logo will not upload. Neither tool accepts an image as the watermark; both are text only. For a logo, LibreOffice Draw or an image editor is the way.

Questions people ask

Is the marked file much bigger? For a PDF, barely — vector text adds a few kilobytes. For photos, output is PNG or WebP if that is what came in, otherwise JPEG at high quality, so size depends mainly on the original.

Does it change the original file? No. You download a new copy and keep an unmarked original for yourself.

Can I watermark several photos at once? Yes, the image tool takes a batch and applies the same settings to all of them, scaled to each image.

Does the watermark print? Yes, including the opacity. A very light mark can vanish on a low-quality print — test one page before sending a hundred to the printer.

Should I number the pages too? If the document will be discussed in a meeting, add page numbers after watermarking, so the two layers do not collide over the same corner.

Next steps

• Flatten a filled form before you mark it: Flatten PDF
• Add real protection on top: Protect PDF, and the guide How to password-protect a PDF
• Mark a batch of pictures instead: Add Watermark to Image