Watermark a batch of images

How placement and size are calculated

A watermark specified in raw pixels looks oversized on a small thumbnail and disappears on a large photo. This tool uses proportions instead. The font size defaults to 4.5% of the image width, and a logo defaults to 15% of the width. The edge margin is 3% of the shortest side, which keeps the visual gap even on landscape and portrait files.

Corner positions are measured from the rotated bounds of the mark, not from an invisible unrotated box. That matters as soon as you add rotation: a 20-degree text mark needs a wider footprint, and a naive corner calculation would clip it. The tool measures the text with the Canvas text metrics, computes the rotated bounding box, and clamps the center so the mark stays inside the image. Custom X and Y values use percentages of image width and height and refer to the center of the watermark, so 50/50 is the middle of the frame.

Text watermarks that remain readable

Text is drawn with the operating system's interface font, which means the exact letterforms can vary slightly between Windows, macOS and Android. The weight and size are consistent, but a word that fits on one computer may be a little wider on another. The tool auto-shrinks a mark that would overflow the image, and it stops with an error if the text cannot fit at a usable size. Keep a copyright line short rather than relying on tiny type.

Contrast is the difference between a watermark that reads and one that looks like a smudge. White text over a dark photograph works without help; white text over a bright sky needs the dark background box. The box is measured from the same text metrics, including padding, so it does not cut through ascenders or descenders. Opacity controls both the text and the box together. Lower opacity is easier on the eye, but it also makes the mark easier to erase.

Logo watermarks and transparency

Use a PNG with a transparent background for a logo. A JPEG logo carries a solid rectangle of background color, which will be obvious over every photo. If the logo is white-on-transparent, it will vanish against pale areas unless you add a dark backing in the logo file itself or place it over a consistently dark part of the image. The scale field sets the logo width as a percentage of the source width; the height follows the logo's own proportions.

A watermark is usually easier to remove than people expect, so do not treat it as security. Keep the mark visible enough to identify the owner, and avoid placing it only in a corner where a crop would remove it. For a portfolio preview, one mark near the center or a repeated pattern is harder to crop out than a single small corner stamp. The full-resolution original should stay off public pages in either case.

Batching, output formats and limits

Every file in the queue receives the settings that are on screen when that file starts processing. If you change the text or position halfway through a batch, the remaining files use the new values while finished rows keep their original result. Mixed portrait and landscape images are fine: the size is relative to each image, so the mark stays proportionate rather than identical in pixels.

Keeping the source format is the safest default. Choose PNG when the source or logo needs transparency, and JPEG when you want a small file for a photo. JPEG cannot store transparency, so transparent areas are flattened against white. Animated GIFs are reduced to their first frame, as with any canvas-based editor. The limit is 50 MB per file and 40 megapixels, which keeps decoding within the memory available to a typical phone browser. For larger scans, resize a copy first with the image resizer.