Does a watermark actually protect your photos?
Partly. A watermark is a deterrent and an attribution, not a lock. Understanding which of those you need decides everything about how you place one.
Last reviewed
What a watermark actually does
Three things, and it is worth separating them because they call for different marks.
It deters casual reuse. Most unauthorised use is not malicious; it is somebody grabbing an image because it was there and easy. A visible mark makes that slightly awkward, and slightly awkward is enough to stop a large share of it.
It attributes. When an image travels — and images travel — the mark travels with it. Anyone who sees it knows whose work it is and where to find more. For a photographer or an illustrator this is often worth more than the deterrence.
It marks a proof. A heavily marked preview lets a client see the work without having a usable file. This is the one case where a watermark genuinely functions as a control rather than as a signal.
What it does not do
It does not prevent copying. Anything displayed can be screenshotted.
It does not survive a determined person. A corner mark is cropped off in one action. A semi-transparent mark over a busy background can be painted out by anyone with basic editing skills and some patience, and — worth being straightforward about — modern content-aware fill and inpainting tools have made this considerably easier than it was.
It does not establish copyright. You already hold that from the moment you make the image, in essentially every country. A watermark is evidence of a claim and of bad faith by an infringer who removed it; it is not the source of the right.
Anyone selling watermarking as protection is overselling it. It raises the cost of casual reuse, and that is a real and useful effect.
The trade-off is one axis, not two
Visibility and removability are the same dial. A mark unobtrusive enough to leave a photograph looking good is unobtrusive enough to crop, clone or paint out. A mark that genuinely cannot be removed cleanly — large, tiled, across the middle of the subject — ruins the photograph as a photograph.
So decide which job the mark is doing before touching any slider:
Attribution and light deterrence — a corner, small, forty to sixty per cent opacity, outlined so it reads on any background. Accept that it can be cropped. That is fine, because someone willing to crop it was never going to credit you anyway.
A proof copy — tiled diagonally across the whole frame, clearly legible, unapologetically in the way. Nobody is going to publish it, which is the point.
There is not much of a useful middle. A large, faint mark across the centre is the worst of both: annoying to look at and, because there is so much surrounding detail to reconstruct from, often easier to remove than a solid one.
Practical placement
Keep it away from the very edge. A mark tight to the corner is trimmed off by a two per cent crop that nobody would even notice. A margin of four or five per cent means removing it costs visible picture.
Outline it. White text disappears against sky and black text disappears in shadow, and a photograph usually contains both. An outline is what makes one setting readable across a whole set of images.
Size it as a proportion. A mark specified in pixels is illegible on a 6000-pixel photograph and covers a third of a thumbnail. As a percentage of the short edge it stays the same relative size on everything, which is what you were judging by eye anyway.
Do not put it over the subject unless you mean to. Over a face it destroys the picture; over a corner of sky it is nearly free.
The thing most people forget
Watermark the version you publish, not your original. Keep the clean file. This sounds obvious and is the most common regret in the whole exercise — a marked copy is not reversible, and the original is what you will need when someone licenses the image.
FileBelow never modifies the file you already have; it hands you a second, marked copy. That is a property of how it works rather than a promise, but it only helps if you keep track of which file is which.
Two things worth doing alongside it
Fill in the copyright field. EXIF has artist and copyright fields that almost nobody uses. They are trivially removable, so they are not protection — but they survive an innocent copy, and they cost nothing. What EXIF data is covers what else is in there.
Publish at a sensible size. A 6000-pixel original posted publicly is a print-quality file given away. Something around 1500 to 2000 pixels on the long edge looks perfect on any screen and is not much use to anyone wanting to print it. The resizer does that, and the watermark tool will apply the mark and a size limit in the same pass.
Need to actually do this?
Text or a logo, sized as a proportion of the image so one setting works at any resolution. Free, no sign-up, and the image is processed on your own device.
Add a watermark to a photoOr browse every explanation on this site, and the requirements published by named destinations.