You want to post a photo and keep your name on it. Then you have to decide three things — where, how large, how strong — and because each one fights the others, 'somewhere in the middle' rarely lands well.
Decide what it is for first
Watermarks serve one of two purposes, and the two point in opposite directions.
- **Attribution** — the point is that people know who made it. Not spoiling the photo matters more, so: corner, small, faint.
- **Deterrence** — the point is to make the image annoying to reuse. It has to be hard to remove, so it has to cover the frame.
Portfolio and social posts are almost always the first case. Paid previews and client proofs — images that should not be usable as-is — are the second. Pick a compromise without deciding, and you get something that looks bad and stops nobody.
What each placement costs you
| Placement | Damage to the photo | Effort to remove | Suits |
|---|---|---|---|
| Bottom-right corner | Almost none | Trivial — one crop | Attribution, portfolios |
| Centred on an edge | Low | Easy | Thumbnails with fixed dimensions |
| Large, centred | High | Moderate — tedious to paint out | Proofs and previews |
| Tiled diagonally | Medium | Hard — it is everywhere | Deterrence |
15 to 30 percent opacity is the working range
Too faint and it vanishes on bright photos; too strong and it covers them. For white text, somewhere between 15 and 30 percent usually sits right. How it reads depends heavily on the underlying image, so use dark text on light photos and light text on dark ones rather than one setting for everything.
Adding a soft shadow or a thin outline to the text keeps it legible over both. That is less work than tuning the colour photo by photo.
Size it against the image width
Fix the size in pixels and it turns into dust on a large photo and swallows a small one. Expressing it as a share of the image width stays consistent. For a corner mark, 15 to 25 percent of the width is usually legible without being intrusive.
This shows up most when you batch a folder: mix portrait and landscape shots and a pixel-fixed watermark looks like a different size on every frame.
What a watermark cannot do
Honestly: it will not stop someone determined to remove it. A corner mark is one crop away, and a watermark over a flat background disappears under any healing tool. Tiling only raises the effort.
Which is why a watermark is better understood as **a signature rather than a lock**. Seen that way, there is far less reason to wreck the photo with a heavy one. If you want to protect something that actually causes harm when leaked, look at the EXIF data instead — GPS coordinates from a photo taken at home are the real exposure.