FineDocs
English

Watermark a photo with your own text

Stamp your name, a URL or a notice across a picture, free and with no signup. Choose the corner, the size, the angle and how faint it sits.

Drop files here (up to 40)

or paste from the clipboard

Preview

Choose an image and the result appears here as you change the options.

Shows the framing and geometry exactly. Compression quality is not simulated.

Options

Short works best. A caption longer than the picture is wide has to be set so small that nobody reads it anyway.

A corner is a signature. Tiled is a deterrent — it covers everything, so no single crop removes it.

%

A percentage of the shorter side, not a pixel height, so one setting looks the same on a small screenshot and a photo from a phone.

%

Around half is readable without fighting the picture. Below a quarter it survives a screenshot but not a determined edit.

°

Ignored when tiled, which always sits at a fixed diagonal.

#ffffff

White reads over most photographs; black reads better over a pale document scan.

%

Only applies to JPEG inputs. PNG and WebP come back as PNG with no quality spent.

Results appear here. Nothing is uploaded.

    How it works

    The picture is drawn onto a canvas at its own size, and the caption is drawn on top of it twice: once as a dark outline and once as the fill you chose. The two passes are what keeps white text legible where it crosses a bright sky, which a single flat fill never manages. Size, margin and outline thickness are all fractions of the shorter side, so the same settings produce the same look across a batch of files that are not the same size.

    What to watch out for

    A watermark is a deterrent, not protection

    Anything drawn onto a picture can be painted out again, and modern editors do it well enough that a corner signature takes a determined person about a minute. Tiling across the whole frame makes it tedious rather than impossible. If an image must not circulate, the answer is not to publish it at full resolution.

    The caption is drawn in the system typeface

    A worker has no page to inherit fonts from and cannot load one from your computer, so the text is set in whatever the platform calls its interface font. That means a Mac and a Windows machine produce slightly different letterforms from the same settings, which matters if you are matching an existing brand mark exactly.

    JPEG inputs are re-encoded

    The picture has to be decoded to be drawn on, so a JPEG is compressed again on the way out and spends a little quality doing it. At the default setting the difference is hard to see, but watermarking an already heavily compressed file twice will show. PNG and WebP inputs are written back as PNG and lose nothing.

    Limits

    Files up to about 50 MB work reliably. Beyond that, the ceiling is your device's memory rather than any rule we impose — the work happens on your own machine. Tested in Chrome 140, Firefox 143, Safari 18.

    What a watermark is actually for

    There are two reasons to write across a picture and they want opposite settings.

    The first is attribution. A photographer putting a name and a site in the bottom corner is not trying to stop anyone from taking the image — they are making sure that when it travels, the credit travels with it. That wants small, faint and out of the way, in a corner where it will not sit on the subject’s face.

    The second is deterrence. A proof sent to a client, a document marked DRAFT, a portfolio shown before it is paid for: here the point is that the file is unpleasant to use as it stands. That wants the tiled option, which lays the text diagonally across the entire frame so no single crop is left clean.

    Why the size is a percentage

    Every setting here is proportional to the image rather than fixed in pixels, and that decision shows up the moment you drop in more than one file.

    A caption set at twenty-four pixels is a bold statement across a screenshot and nearly invisible on a twelve-megapixel photograph. Given a batch of both, a fixed size produces two completely different results. Six percent of the shorter side looks like six percent of the shorter side on every file you give it.

    Everything happens on your machine

    The picture is decoded, drawn and encoded inside a Web Worker in this browser tab. Nothing is uploaded, which means there is no queue, no daily limit and no copy of your photographs on hardware belonging to somebody else — a point worth making about a tool people reach for precisely when an image is not meant to circulate freely.

    Forty files can go through at once. The result panel shows the first one so you can check the placement before downloading the rest.

    Your files never leave your browser. Last updated 2026-08-07.