Add a Watermark to an Image — Free

What a watermark does and does not protect

Worth being clear about before you spend time on the design: a visible watermark is a deterrent, not protection. Anyone determined to use your picture can crop a corner mark away, clone-stamp a small one out, or run it through one of the AI removal tools that have become genuinely good at exactly this. No arrangement of text prevents that.

What a watermark does do, and does well:

• Stops casual reuse. Most unlicensed use is not theft by a determined adversary — it is somebody dragging an image into a deck or a blog post because it was there. A mark makes that feel wrong, which is usually enough.
• Carries the source with the picture. When an image is reposted, the credit usually is not. A URL in the corner travels with it.
• Marks proofs as proofs. This is the strongest use: a client sees the photograph, and the mark makes plain that the clean file comes after payment.

What it cannot do is prove ownership after the fact. If that is the goal, keep the original raw files and their metadata, and register the work where registration matters in your jurisdiction. A watermark is a sign on the gate, not the lock.

Where to put it, and how visible to make it

The choice is a straight trade between how hard the mark is to remove and how much it spoils the picture.

Tiled across the whole picture

Hardest to remove, because there is no clean region left to clone from and cropping loses the picture along with the mark. It is also the most intrusive, which is why it belongs on proofs and preview copies rather than anything you want admired. Keep the opacity low — around 20–30% is legible without drowning the image, and a low, repeated mark is harder to paint out than a single bold one.

A corner

Least intrusive and the usual choice for work you are publishing rather than proofing. It is also the easiest to defeat: a crop takes it off and the picture survives. Bottom-right is conventional; bottom-left is worth choosing if your subject tends to sit on the right. Make it large enough to read after the platform has resized it — social feeds serve images far smaller than you uploaded.

Centred

A compromise: harder to crop than a corner, less ruinous than a tile. It works best over a busy area, where it is awkward to reconstruct, and worst over a plain sky, where any editor removes it in seconds.

Whichever you choose, the text is drawn white with a dark outline behind it. That is not decoration — white alone disappears against a bright sky or a sandy beach, and the outline is what keeps it readable on both.

Invisible watermarks are a different thing

This tool draws a mark you can see. There is a different technology — variously called invisible, forensic or steganographic watermarking — that hides an identifier in the pixel data itself, where it survives resizing and re-compression and is designed to be recoverable later even from a copy.

We do not do it, and would rather say so than let the word "watermark" imply it. It needs a matching detector to be worth anything, which means a service that keeps a record — the opposite of a page that sends nothing anywhere. Agencies and stock libraries use it precisely because they run both halves.

If you need the visible mark, this page is the whole job. If you need provable attribution, you need a service built around that, and a visible mark on top of it as the part people can actually see.

Capabilities at a glance

What this tool does and does not do, stated plainly. The “No” rows are the point, not an omission.