Remove a Watermark From a PDF

What "removable" means here

A PDF is a set of objects, not a picture of a page. Whether a watermark can be removed cleanly comes down to one question: is the mark still a separate object, or has it been merged into the page content?

Still separate — this tool can delete it:

• Watermark and stamp annotations. Added by Acrobat and most PDF editors as an annotation layer sitting above the page. Deleting the annotation leaves the page itself untouched.
• Optional content groups. The PDF name for layers. A watermark on its own layer can be removed along with the marked-content section that draws it.
• Repeated text objects. A "DRAFT" or "CONFIDENTIAL" text block drawn at the same coordinates on every page. It is part of the content stream, but it is identifiable precisely because it repeats, and the block that draws it can be cut out.

Merged into the page — nothing can honestly remove it:

• Flattened watermarks. Many tools "flatten" a document on export, which merges every layer into one. After that the mark is ordinary page content with no marker saying it was ever separate.
• Rasterised pages. If the page is one big image, the watermark is pixels in that image.
• Watermarks that were in the original. A scan of a sheet that was already stamped, or a photo of a marked document. The mark was there before the PDF existed.

For that second group, what other sites call watermark removal is one of two things: painting over the area and guessing what was underneath, or cropping. Both change the document in ways you cannot verify, which is worse than a clear "no" when the document is a contract or a licence. This tool does neither.

If your watermark is the kind that cannot be removed

There are three honest options and it is worth knowing which one you are in before spending an afternoon.

Ask for a clean copy. Usually the fastest route and usually overlooked. Draft watermarks in particular exist because a final version also exists.

Recover the text and rebuild. OCR reads the words off the page, watermark included, and gives you text you can edit. You lose the original layout entirely, so this suits documents where the content matters and the formatting does not.

Crop it out. Only works when the mark sits in a margin rather than across the middle. Export the pages as images, crop, and rebuild with image to PDF. The result is no longer searchable text.

What none of these do is restore what was under a diagonal stamp across the body of the page. That information is not in the file.

How the scan works

1. The file is parsed in your browser with pdf-lib. Nothing is uploaded.
2. Annotations are listed per page, and any with a Watermark or Stamp subtype are flagged.
3. Optional content groups are read from the document catalogue and matched to the marked-content sections that draw them.
4. Content streams are decompressed and scanned for text blocks. A block whose text and position repeat across most pages is flagged as a likely watermark — which is also, deliberately, how a running header looks, so nothing is removed until you tick it.
5. You choose what to remove, see page one rendered before and after, and download the result only if it looks right.

The output is a new PDF written from the same objects minus the ones you removed. Text stays selectable, pages are not re-rendered, and nothing is re-compressed, so nothing else in the document degrades.

Next steps

• Add a watermark to a PDF — the other direction, if you are the one marking a document.
• OCR — recover the text when the watermark is part of the page image.
• PDF metadata — the author, software and timestamps a watermark removal does not touch, and which often say more than the watermark did.

Next steps