A corner watermark is removed by cropping the corner
That is the sentence missing from every other watermarking tool, and it takes about four seconds and no skill at all.
The corner is what almost everybody asks for. It looks tidiest, it stays out of the picture, and it is the arrangement least likely to survive contact with somebody who wants your image.
A tiled watermark, repeated across the whole frame, has no such weakness: there is no crop that keeps a usable picture and loses the mark. It is also uglier. Both are here, and the panel tells you which one you have chosen and what it means.
What a watermark stops, and what it does not
It does not prevent theft. Somebody determined will clone it out, crop it, or run an inpainting model over it, and those tools are free and getting better.
What it does is two things, both genuinely useful.
It stops casual reuse. Most images are not stolen by determined people, they are taken by somebody in a hurry who needed a picture for a listing, a post or a slide. A visible mark makes that not worth the bother, and they go and find another picture.
It carries attribution. Images travel a long way from where they started, and a mark is the only part of the file that survives being screenshotted, reposted and re-uploaded. Anybody who wants to license it, credit it or commission you can see whose it is.
Sold as protection it is a lie. Used for those two things it does the job.
Size in per cent, never in pixels
A watermark set at 40 pixels is right on exactly one image and wrong on every other one you own.
On a 6000 pixel photograph off a modern camera it is a speck nobody will notice. On a 400 pixel thumbnail it covers a third of the frame. This is why batch watermarking most often looks amateurish.
So the size here is a percentage of the image's smaller side. Smaller rather than width, because sizing off the width makes a mark that looks right in landscape and oversized the moment you turn the picture on its end.
Opacity, and the thing that eats it
Between 30 and 60 per cent suits most photographs. The temptation is always to go lower, because a faint mark looks more professional on your screen.
What that misses is recompression. Every messaging app, social platform and email client re-encodes an image, and a very faint watermark is exactly the sort of low-contrast detail JPEG compression throws away first. A mark that reads at 10 per cent on your monitor can be effectively gone by the time it has been through a couple of apps.
Below about 15 per cent the tool says so rather than letting you export something that does nothing.
Text with a shadow beats a logo on unpredictable images
A logo is the obvious choice and it has one persistent problem: it is one colour and photographs are not.
A white logo disappears on a bright sky. A dark one disappears on a shadow. If your images are consistent, that is fine and a logo looks better. If they are not, text with a drop shadow stays legible on both, because the shadow provides the contrast the background is not.
If you do use a logo, use a PNG with a transparent background. A JPEG logo brings a white rectangle with it and nothing here can remove that, because as far as the file is concerned the white is part of the picture.
Nothing uploaded, nothing added
Every image is decoded, marked and saved on your own device, including a batch of a hundred. There is no queue, no size cap and no account.
And nothing of ours goes on your picture. No badge, no corner credit, no held-back resolution. A watermarking tool that watermarks your image with its own name would be an odd thing to build, and yet.
Common questions
Does a watermark stop people stealing my photos?
No, and anybody telling you otherwise is selling something. A corner watermark is removed by cropping the corner. A tiled one can be cloned out or run through an inpainting model. What a watermark does is stop casual reuse, the sort where somebody drags your image into a listing without thinking, and carry your name with the file so whoever sees it downstream knows whose it is. Both are worth having. Neither is protection.
Corner or tiled?
Corner if the point is attribution and you want the picture to look good. Tiled if you mind about reuse, because there is no crop that keeps a usable image and loses the mark. It is uglier in exchange. Proofs and previews sent to clients are the classic case for tiled; a portfolio piece is the classic case for a corner.
What opacity should I use?
Somewhere between 30 and 60 per cent for most images. Below about 15 it is effectively invisible once a messaging app or a social platform has recompressed the picture, which they all do, so it deters nobody and attributes nothing. Above about 70 it stops being a watermark and starts being a graphic on top of your photo.
Why does my watermark look different on different photos?
It should not here, because the size is a percentage of the image rather than a number of pixels. A watermark set in pixels is invisible on a 6000 pixel photo and enormous on a 400 pixel thumbnail. Sizing off the smaller side keeps it proportionate on landscape and portrait alike.
Can I use my logo instead of text?
Yes. Use a PNG with a transparent background: a JPEG logo brings its white rectangle along with it and there is no way to remove that afterwards. If your logo is dark, put it on light areas or turn the opacity up, and if it needs to work on both, text with a shadow is more reliable.
Can I do a whole folder at once?
Yes. Add as many images as you like, set the watermark once, and they all come back in a zip. The size stays proportionate on each one, so a mix of landscape, portrait, large and small all get a mark that looks deliberate rather than one that is right on one file and wrong on the rest.
Is anything added by you?
No. There is no badge, no corner logo, no reduced resolution and no account. A watermarking tool that puts its own watermark on your picture would be a strange thing to build.