Image Text Watermark

Renders any text directly onto your image in a canvas, so the result is a flat PNG with the watermark baked in. Pick one of nine anchor positions, or switch to tile mode to repeat the text across the whole image (common for proof watermarks). Size, opacity, rotation, colour, and drop shadow are all live — what you see in the preview is the file you'll download.

Everything happens in the browser: the image is decoded, drawn to a canvas at its natural pixel size, and re-encoded locally. Nothing is uploaded. The consequence of that round trip is that the output is always PNG, whatever you put in, so a JPEG photograph will usually come back several times larger than the original. Re-encoding also discards everything that was not pixels — EXIF, ICC profiles, GPS coordinates, capture timestamps and any embedded colour profile are gone from the download, which is useful when you wanted them stripped and surprising when you did not.

Text is drawn in the browser's system-ui stack, so the exact glyph shapes depend on the operating system rendering the page. The same settings produce a slightly different width on Windows, macOS and Linux, and a font size is measured in canvas pixels rather than as a share of the image, which means one value that looks right on a 4000px photo will be unreadable on a 600px thumbnail. Tile mode spaces repetitions from the measured text width, so longer text automatically spreads further apart.

What a visible watermark cannot do is protect the image. Anyone can crop a corner mark away, and modern inpainting removes even a tiled one with modest effort. It marks provenance and discourages casual reuse; it is not a rights-management measure, and for tracking a leak an invisible method is the appropriate tool.