Image Watermark Tool — your mark, added in your own browser
This image watermark tool draws a line of text, a logo file, or both onto a copy of a photo you already own. You load the picture from your device, type your handle or pick a logo, place the mark with a nine-point grid or by dragging it straight across the preview, then download the composited result. The photo is decoded and redrawn by your own browser on a <canvas> element, and no version of it is ever uploaded — there is no server here to upload it to.
What a watermark actually does
A visible watermark is a deterrent and an attribution mark. It makes casual reuse more awkward, and it keeps your name attached to the picture as it travels, so anyone who comes across it can see where it came from. That is a real and useful job, and it is the only job the mark does. A corner mark can be cropped away in seconds; a mark anywhere on the frame can be painted over by someone willing to spend the time. Your copyright in the photograph exists whether or not you stamp it — the mark neither creates that right nor replaces it.
Why the size is a percentage, not a pixel count
Every size here — the text, the logo, the inset from the edge, the gap between tiles — is a percentage of the image's shorter side. That is not a stylistic choice. 48px of text is a confident signature across a 640 × 480 thumbnail and a barely visible speck on a 6000 × 4000 camera original: one slider value, two completely different pictures, with nothing on screen to tell you which one you were about to download. The preview is also a scaled-down copy, so a pixel figure that looked right on screen would not be the figure applied to the full-resolution file.
Percentages remove that whole class of surprise. Four per cent is four per cent of the frame on any image, the preview and the export are drawn by one function that differs only in its scale factor, and the line of figures under the canvas tells you what your percentage works out to in real pixels for the photo you actually loaded.
Contrast is what makes a mark readable
A white mark disappears into a snow scene; a black one disappears into a night shot. This tool will not sample your image and guess, because a photograph has no single background colour and a guess that is wrong some of the time is worse than no guess — you would stop checking the preview. Instead you pick the light or dark variant yourself, and add a legibility style: an outline strokes a contrasting edge around the mark, a shadow drops a soft blur behind it, and a plate puts a translucent rounded rectangle underneath, which reads most clearly and intrudes most. Opacity is a separate slider, and it applies to the mark alone — never to your photograph.
One mark or many
A single mark in a corner stays out of the picture and is the easiest thing in the world to crop off. A tiled lattice repeats the mark across the whole frame, usually turned about -30°, which is far more work to paint out and considerably more intrusive to look at. Neither setting is correct; they trade the same difference in opposite directions, and the preview is there so you can judge it. Press and hold the before/after control to see the untouched image and decide how much the mark is costing you.
Choosing an export format
PNG and WebP keep an alpha channel; JPEG cannot, so a base image with transparent areas is flattened to a solid colour — black in some browsers, white in others — and the tool warns you when that combination is selected. For an ordinary opaque photo all three are fine, and JPEG or WebP will give you a much smaller file. The quality slider appears only for JPEG and WebP, because PNG is lossless and ignores the quality argument entirely.