Photographers, illustrators and shop owners watermark images for the same reason artists have always signed work: so that a picture separated from its source still says where it came from. Once an image is on the internet it will be screenshotted, reposted and scraped, and a visible mark is the only attribution that travels with it.
The mark is drawn directly onto the image, rotated thirty degrees across the centre, at a size proportional to the picture — one tenth of its shorter edge — so it reads consistently whether the file is a thumbnail or a full-resolution export. Because it is rendered through the browser’s text engine rather than a fixed font set, any script works: Korean, Japanese, Chinese, Cyrillic and Arabic all display correctly.
The watermark is composited in your browser. Unpublished work is marked without being uploaded to a service first — which rather defeats the purpose otherwise.
-edited.| Opacity | Reads as | Use when |
|---|---|---|
| 10% | A faint signature | Portfolio work where the image must look clean |
| 20% | Clearly present, not intrusive | The usual choice for social and client previews |
| 40% | Impossible to ignore | Proofs and samples that must not be used as finals |
There is a genuine tension here. A faint mark keeps the image attractive but is easy to clone out; a heavy one is hard to remove but makes the image unpleasant to look at. For client proofs, 40% is correct precisely because the discomfort is the point. For a portfolio, 10% to 20% is the honest balance.
It is worth being clear-eyed. A determined person with an image editor can remove a diagonal text watermark, and automated tools exist that do a passable job. What a watermark reliably does is stop casual, thoughtless reuse — the blog post that grabs the first image in a search, the marketplace listing that lifts a competitor’s product shot — and it means that when your work does travel, your name travels with it.
If an image genuinely must not be reused, the only reliable protection is not publishing it at full resolution. Watermark the preview and keep the original.
Yes. The text is rendered through your browser’s font stack, so Korean, Japanese, Chinese, Cyrillic, Greek and Arabic all display correctly.
Not currently. The mark is centred and rotated thirty degrees, which is the hardest placement to crop out and works on any aspect ratio.
It scales with the image — the text height is one tenth of the shorter edge, with a minimum of 24 pixels. That keeps it proportionally consistent across a batch of differently sized files.
With effort, yes. A visible watermark deters casual reuse and carries attribution; it is not a security measure. For images that must not be reused, publish only a low-resolution watermarked version.
Yes. The same text and opacity are applied to every image in the batch, each scaled to its own dimensions.
No. The text is drawn onto a canvas in your browser and the marked file is written straight to your downloads.