Skip to content
ToolBoxGeniehome

Image Watermark

Image Tools · Added

Add your name, your site address or a status such as DRAFT across a photograph, either as a single mark in one corner or repeated over the whole frame. Everything happens on a canvas in this page — the files are read from your device and never uploaded — which is what makes this safe to use on an unreleased photograph or a scanned document.

Or drop files onto this box.

JPG, PNG, WebP, GIF, BMP or AVIF. Up to 10 files, 30.00 MB each. They are read from your device and never uploaded.

Your name, a site address, or a status such as DRAFT or CONFIDENTIAL.

Rendered with the fonts already on this device.

Where the caption sits.

The outline is set to whichever contrasts.

A share of the image's shorter side, so a batch of mixed sizes comes out consistent.

Lower is more discreet and easier to crop out.

Turned about the caption's own centre.

Distance from the edge, ignored when the position is centred.

GIF and BMP have no encoder in the browser and come out as PNG.

PNG is lossless, so quality does not apply.

How to use the image watermark

  1. 1Choose one or more images, or drop them onto the box. Up to ten at a time.
  2. 2Type the watermark text. The size is set as a percentage of each image's shorter side, so a mixed batch comes out looking consistent rather than tiny on some files and enormous on others.
  3. 3Pick a position and adjust opacity, or turn on "Repeat across the whole image" for a tiled mark that is much harder to crop away.
  4. 4Rotate the text if you want it running diagonally, and leave the outline on if the mark has to sit over both light and dark areas.
  5. 5Check the preview, then download each image. Every setting reapplies to the whole batch automatically.

Examples

A discreet credit on a photograph

Input
"© Studio Name" · bottom right · 5% size · 45% opacity · 4% margin
Result
A small semi-transparent caption in the corner — 5% of a 3000px short side is a 150px cap height

Corner marks are the least intrusive and the easiest to crop off. They deter casual reuse and identify the source; they do not protect anything.

A document marked as a draft

Input
"DRAFT — NOT FOR CIRCULATION" · centre · 8% size · 30% opacity · rotated -30°
Result
A large diagonal mark across the middle of the page

Rotation is applied about the caption's own centre, so turning it does not also drag it away from where you positioned it.

A tiled mark on a proof

Input
"proofs.example.com" · tiling on · 40% spacing · 25% opacity
Result
The address repeated in a grid across the whole frame, clipped at every edge

The grid is centred on the image so the overhang matches on opposite sides, which is what stops a tiled mark reading as a block of text pasted over the middle.

About the image watermark

Corner mark or tiled: choosing what you are protecting against

The two placements answer different worries. A corner mark is about attribution: it keeps your name on the image as it travels, and it stays out of the way of the picture. A tiled mark is about preventing use: it makes the image unattractive to lift, because removing the pattern means reconstructing whatever is underneath it across the whole frame.

The cost is proportional. A corner mark is nearly free — the photograph still works as a photograph — and nearly free to remove. A tiled mark meaningfully damages the image as a viewing experience, which is exactly why it works on proofs, previews and anything sent to a client before payment.

Opacity is the dial between the two extremes rather than a compromise between them. A tiled mark at 15% is often enough to make an image unusable commercially while still letting a client judge the composition, and that is usually the setting a proof wants.

Why sizing by percentage rather than pixels

A watermark specified in pixels is wrong on almost every image except the one it was designed on. Twenty-four pixels is a bold statement on a 400px thumbnail and an illegible speck on a 6000px camera file — and a batch of photos straight off a phone, a camera and a screenshot tool will span exactly that range.

Measuring against the shorter side rather than the width is a deliberate second choice. The width alone would make the same setting produce a much larger caption on a panorama than on a portrait shot of the same subject, because the short side is what actually limits how much room the mark has in either orientation.

The same reasoning applies to the margin, which is also a percentage. A fixed 20px inset looks generous on a small image and hairline on a large one; a percentage keeps the mark sitting the same visual distance from the edge whatever it is applied to.

Keeping text legible over an unpredictable background

A white caption disappears on a snow scene and a black one disappears on a night shot, and no tool can know in advance which it has been handed. Worse, a mark long enough to be useful frequently straddles both — the last three characters land on a bright sky while the first three sit on a dark treeline.

The outline option is the answer to that, and it is on by default. The text is stroked in whichever of black or white contrasts with the fill colour you picked, so a light caption gains a dark edge and a dark one gains a light edge. It is a small effect at full opacity and the difference between readable and invisible at low opacity, which is where most watermarks want to sit.

Frequently asked questions

Are my photos uploaded anywhere?
No. The file is read into the page, drawn onto a canvas with the text on top, and re-encoded — all inside your browser. Nothing is sent to a server, which is the reason this is usable on work that has not been published yet. You can confirm it by opening the network tab of your browser's developer tools and watching it stay empty while you work.
Can I use an image or logo as the watermark instead of text?
Not in this tool — it stamps text only. A logo watermark needs a second file, its own scaling and transparency handling, and a different set of controls, and doing it properly is a separate tool rather than a checkbox on this one. For now, a text mark in a distinctive typeface covers the common case of identifying the source.
Does a watermark actually protect my work?
It deters, it does not protect. A visible mark makes casual copying obvious and keeps your name attached to an image as it is shared on, which is genuinely useful. It is not copy protection: a corner mark can be cropped in seconds and a semi-transparent one can be painted out with modest effort. Tiling across the picture raises that effort considerably, at the cost of a busier image — which is the trade this tool leaves to you.
Does the watermarked copy keep my copyright metadata?
No, and on this tool in particular that is worth knowing. The output is drawn from pixels onto a fresh canvas, so IPTC and EXIF blocks do not survive — including any creator or rights field set in your photo software. On a copyright tool the implication is pointed: the visible mark becomes the only attribution travelling with the copy you download. Treat your original as the master, and re-apply embedded metadata afterwards if your workflow relies on it.
Why does my watermark look different across a batch?
It should not, and the size setting is the reason. The font size is calculated as a percentage of each image's shorter side rather than as a fixed number of pixels, so the same setting produces a proportionally identical mark on a 600px thumbnail and a 6000px camera file. If one image still looks different, check whether it is a very different aspect ratio — a panorama has far more room along its long edge than the short-side measurement accounts for.
Which output format should I choose?
Keep the original unless you have a reason not to. JPEG is right for photographs and keeps files small; PNG is lossless and right for screenshots or anything with sharp text, at a considerably larger size; WebP is usually smaller than both at the same visible quality. GIF and BMP have no encoder in the browser, so those come out as PNG — the tool tells you when that happens rather than producing a file with a misleading extension.

Related tools