A watermark is a signature, not a lock
It is worth being honest about what this does. A corner watermark takes about four seconds to crop out. A tiled one takes longer, and modern removal tools handle even that with some effort. If your goal is to make theft impossible, no watermark will get you there.
What a watermark actually does is travel with the image. When a photo is reposted, screenshotted, scraped into a feed and reposted again, your name goes with it. Most of the value is attribution, not prevention — and attribution is genuinely worth having, because the alternative is your work circulating with no idea who made it.
Corner or tiled: two different jobs
| Corner | Tiled | |
|---|---|---|
| For | Published work | Proofs and previews |
| Goal | Credit without spoiling the image | Make the image unusable until paid for |
| Opacity | 40–60% | 20–35% |
| Removable | Easily, by cropping | Only with real effort |
The mistake is using the wrong one. Tiling a photo you want people to enjoy and share ruins it, and nobody shares a ruined photo. Putting a discreet corner mark on a client proof means the client has no reason to pay you — they can just use the proof. Match the treatment to the purpose.
The tiled pattern here is rotated and offset row by row rather than laid out on a straight grid. A straight grid can be cropped between the marks on a large image; a rotated, staggered one leaves no clean rectangle.
Why your white watermark disappears
This is the most common complaint, and it has an obvious cause: a white mark on a white sky is invisible, and a black mark on a dark corner is invisible too. There is no single colour that works on every photograph, and picking per-image defeats the point of batch processing.
The fix is the shadow option, which is on by default. It draws a soft dark edge behind the mark, so white text stays readable over a bright sky and still reads over dark foliage. It is the same trick subtitles use, for the same reason.
If you are using a logo, build the contrast into the file: a white logo with a subtle dark outline, exported as a transparent PNG, behaves far better across a mixed set of photographs than a flat single-colour mark.
Size is a percentage, not a pixel count
The size setting is a percentage of each image's width rather than a fixed number of pixels, and that matters the moment you process more than one file. A 200-pixel watermark is enormous on a 600-pixel thumbnail and nearly invisible on a 6000-pixel export. As a percentage, every image in the batch gets a mark of the same visual weight.
Around 12–16% suits most corner marks. Below 8% it becomes decorative rather than legible; above 25% it starts competing with the photograph.
What the watermark does and does not survive
The mark is drawn into the pixels, so it survives copying, re-saving, format conversion and compression. It does not survive being cropped off or painted over, and it will not help you in a dispute.
If you need to prove ownership rather than display credit, what matters is your original files: the full-resolution raw or camera file, with its untouched capture metadata, in a place with a verifiable date. That is evidence. A watermark is a label.
Batch and privacy
Drop in a whole folder and every image gets the same treatment, then comes back as one ZIP. Because the work happens in your browser rather than on a server, there is no upload step, no file count limit and no queue — which for unpublished client work is worth more than the convenience.