A watermark getting screenshotted and cleaned up by an AI tool is not a sign that a photographer picked the wrong watermark. It is what happens on every platform that relies on the mark itself to do all the protecting, Pixieset included. The fix is not a better watermark. It is a different setup around it.
Why this keeps happening on client-gallery platforms
Pixieset, like most client-gallery tools, is built around delivering full-size, watermarked proofs to a known client who is expected to eventually pay for the real files. That model works well for a wedding client who already has a relationship with the photographer. It breaks down with strangers, which is exactly what a sports photography audience is: hundreds of people who see a full-size preview, have no ongoing relationship with the photographer, and only need one thing, a clean copy, before they lose interest. Independent testing across 324 watermark removal attempts on four different AI models found no watermark design that survived, regardless of opacity, pattern or placement, because AI inpainting reconstructs the covered area from the surrounding pixels rather than trying to "see through" the mark. A subscription-based gallery platform with a strong watermark still loses to that math.
What actually changes the incentive
The photographers who stop losing sales to this are not the ones with the toughest watermark, they are the ones who make paying faster than stealing. Layered protection, not watermark strength alone, is what working photographers land on: keep the watermark as a signal, and put the real friction somewhere the AI tool cannot touch, the price and the speed of the legitimate purchase.
Where Surf Snaps is built differently
Surf Snaps skips the monthly subscription entirely, so there is no plan to justify with heavier features or bigger galleries. Every uploaded photo gets an automatic watermarked preview, and we tested it directly: 50 removal attempts against ChatGPT, Gemini, Flux and other leading models, and 90 percent could not touch it, with another 5 percent producing a damaged, unusable result. The full breakdown is in our watermark testing writeup. On top of that, the unwatermarked full-resolution file unlocks the instant a buyer pays, starting at $8 per photo, so anyone in that remaining sliver who does get through a failed attempt still finds it faster to just pay than to keep retrying.
Discovery matters here too
Pixieset galleries are private, link-based deliveries built for a client who already knows to expect one. A sports photography buyer usually does not have that link, they have a race they ran or a game they played, and no idea who was on the sideline. Surf Snaps organizes around the event instead of the photographer's name, so buyers search by sport, location and date and find their own photo without ever needing a shared link, which is also the setup that makes a fast, low-friction purchase possible in the first place. The fuller comparison is on Surf Snaps vs Pixieset.
Copyright still matters even when a mark gets stripped
Every photo sold through Surf Snaps keeps its copyright with the photographer, regardless of what a buyer does with it afterward. That does not stop a determined bad actor from stripping a watermark, but it is what makes a takedown request possible against anyone who does, which a weaker or stronger watermark alone cannot provide.
Selling what you shoot
Upload your next session to Surf Snaps and see the model built around this exact problem: automatic watermarking, a starting price low enough to beat the AI workaround, and full copyright retained on every file. You keep 90 percent of every sale, paid monthly by bank transfer. Details are on the how it works page and the FAQ. For the full test results, see we tested whether AI can remove the Surf Snaps watermark.
One small step
Move your next session to a platform priced for strangers instead of known clients, and see whether the AI-removal problem quietly stops mattering once paying is the easier option. Create a free account and test it against your next upload.
