Add a text watermark to an image in your browser — one placement or tiled across the whole picture, at your opacity. Never uploaded, processed on-device. Runs entirely in your browser — your numbers never leave this device.
No image loaded yet. Load one, type your watermark, and the stamped file is rendered on this device at the image's full resolution — nothing is uploaded.
Your image is processed entirely in your browser and never uploaded: it is decoded on your own device (with EXIF orientation applied on load, so phone photos are watermarked upright), your text is drawn over it at the size, position and opacity you choose, and the finished image is offered back as a download in the format you pick. Nothing reaches a server, is stored or logged, and the tool works offline once the page has loaded.
There are two placements. A single watermark sits in one position — a corner credit or a centred notice — and is the right choice for attribution that should be seen without dominating the picture. Tiled mode repeats the text across the whole image at the chosen gap between repeats, typically angled, so that any crop of the picture still carries the mark; that is the mode for preview images, proofs and anything that will be shared before it is paid for or published. Opacity is the working trade-off in both modes: around a fifth to a third solid keeps a photograph fully viewable while marking it unmistakably.
The honest scope, as with Watermark PDF: a watermark is a visible mark burned into the downloaded image's pixels, suited to attribution and to discouraging casual reuse. It is not encryption, access control or proof of ownership, and a determined editor can attempt removal from a lightly marked image — tiling and moderate opacity raise that cost, they do not abolish it. The mark is part of the output file permanently, so keep your unmarked original; the tool's job is the marked copy.
No. The image is decoded, drawn on and encoded entirely inside your browser; it never leaves your device, is never stored or logged, and the tool works offline.
It is burned into the pixels of the downloaded file, so there is no layer to switch off. Determined editing can attack any visible mark; tiled placement at moderate opacity makes casual removal impractical because the mark is everywhere, but a watermark is a deterrent and an attribution, not a lock — this page does not claim more for it.
Single for attribution: a photographer's credit, a logo in the corner, a centred status word. Tiled for protection of previews and proofs that may be cropped or screenshotted in pieces — tiling at a regular gap means every piece carries the mark.
Around 20–35% for marks over photographs: clearly visible, clearly a watermark, and the image underneath remains fully viewable. Corner credits that are meant to be read as signatures are usually set higher, 50–70%, because they sit over a small area.
Dimensions never change — the output has exactly the input's pixel width and height. Byte size can move either way with re-encoding and format choice; the exact before and after byte counts are shown, as in the Image Compressor.
More in Image Tools · Browse all tools