MarkText
Coverage6% wide
Opacity45%
Weight—
Stamp a line of text or your own logo onto a picture and drag it exactly where it belongs. The mark is sized as a share of the image width and positioned in percentages, so the same recipe lands identically on a 900 px screenshot and a 6000 px photograph — and tiling covers the whole frame when a corner mark is too easy to crop away.
Step by step
01Add the picture to be marked
Drop it in. The image is decoded locally and the mark is composited onto a canvas in your browser — the original file on disk is untouched.
02Choose text or a logo
Text takes a string, a face — sans, serif, or mono — a size relative to the image, and a colour. Logo takes a second image, which is the right choice when the mark is a brand asset rather than a line of type.
03Place it, angle it, or tile it
Position is a percentage of each axis, so the same recipe lands correctly on images of different sizes. Rotate for a diagonal mark, or tile it across the whole frame when the point is to make the file unusable rather than to sign it.
04Set the opacity you can live with
45% is the default: readable as a claim, quiet enough not to fight the picture. Push it up for proofs meant to be undesirable, down for a signature on a portfolio piece.
Worked example
Everything here is expressed relative to the picture, which is what makes one recipe work across a whole shoot. These are the pixel results of a single set of settings.
Given
One recipe, two canvases
4000x3000 type at 180px, baseline near 3760x2760
reads as a small corner signature
1200x900 type at 54px, baseline near 1128x828
identical proportions, identical lookAbsolute pixel sizes are what make a watermark look right on one export and comically large on the next. Because size and position here are percentages, a recipe you settle on for a hero shot applies unchanged to a thumbnail — and the share link carries exactly that recipe, so a collaborator can mark their own files the same way.
Before you distribute
A watermark deters, it does not protect
Anything visible can be cropped, cloned out, or painted over by someone determined enough. The realistic goals are attribution when an image is shared honestly, and friction when it is not — not prevention.
Corner marks are the easiest to remove
A signature in the bottom right is one crop away from gone. That is a fine trade for a portfolio piece where the mark is a credit; it is the wrong choice for a proof you are sending to a client who has not paid yet.
Tiling trades beauty for actual friction
A repeated mark across the frame cannot be cropped out without destroying the image, which is exactly why proofing services use it. Keep the opacity low enough that the picture can still be judged, and expect it to look like what it is.
Contrast is what makes a mark survive
White type at low opacity disappears into a bright sky and shows clearly on a dark one. If your library is mixed, an angled mark with a colour that differs from the average subject travels better than one tuned to a single photograph.
The export is flattened
The mark is composited into the pixels: there is no layer to toggle off later, and no way back to the clean file from the exported one. Keep your originals — this tool writes a copy and never touches the source.
Nothing is uploaded, which is the point here
Watermarking usually means the image is not public yet — unreleased shoots, client proofs, press material under embargo. Both the picture and the logo stay on your machine; only the recipe travels in a share link.
The judgement call
Placement and opacity should follow the reason you are marking the file at all.
A portfolio image you want credited
Corner, subtle
Low opacity in a corner reads as a signature. It survives honest sharing and does not fight the work.
A client proof before payment
Tiled
The whole point is that the file is unusable as delivered. Tiling cannot be cropped away.
A press or embargoed image
Tiled + high opacity
Deliberately unattractive. It should be obvious in any screenshot that this was not the released asset.
A brand mark on marketing images
Logo, fixed corner
Consistent placement across a set reads as design. Use percentage positioning so mixed sizes stay aligned.
Proving authorship if the file is stolen
Not sufficient
Visible marks come off. Keep the originals, the raw files, and the timestamps — those are the evidence.
Marking a document you will print
Check the contrast
Low-opacity white vanishes on paper. Test at the density the printer will actually produce.
Reference
FAQ
Load the photo, type your text (or load a logo), then drag the mark to where you want it. Set the opacity low enough that the picture reads through — 30–50% is typical for a corner mark — and export. The whole process runs in your browser, and the original file is left untouched.
For a discreet corner credit, 30–50% keeps the mark legible without competing with the image. For a proof or a preview meant to discourage unlicensed use, 50–70% combined with tiling is the usual choice: visible enough to be a deterrent, transparent enough that the client can still judge the picture.
Tile when the point is protection. A single corner mark is removed by cropping in seconds; a tiled grid cannot be cropped away and is tedious to retouch out. Use a single placed mark when the point is attribution or branding, where a repeated stamp would ruin the picture.
Yes. Switch the mark to Logo and choose the file. Transparency is preserved — the logo's own alpha channel is composited onto the picture, and the opacity slider multiplies on top of it. SVG and PNG both work; scale it with the size slider, which measures it as a share of the image width.
The composite itself is lossless: the mark is drawn over the pixels at full resolution and nothing is resampled. Quality only changes at export, when the file is re-encoded. Choose PNG for a lossless result, or keep the JPEG and WebP quality at 90% or above.
No. Both the photograph and the logo are read and composited entirely in your browser. Nothing is transmitted, stored, or logged, and the share link carries only the watermark settings.
The photograph and the logo are both read and composited in your browser. Neither is uploaded, and share links carry only the watermark settings.
Keep going