The text-behind-subject look — huge type that slips behind the person or object in your photo — usually sits behind a paywall or a watermark. Here it's free: an on-device AI model finds the subject, and your text slides in between it and the background.
Type your words, drag them anywhere on the photo, pick a font, size, colour and opacity, and apply. The preview is pixel-identical to the download, and every apply is one undo step.
How the text-behind effect works
The effect is three layers: your original photo at the back, the text in the middle, and a cut-out of the subject pasted on top. The hard part is the cut-out — pixelpix runs a small neural network (U²-Net) in your browser via WebAssembly to separate the subject from the background, so the text genuinely disappears behind a person's shoulder or a building's edge instead of just sitting on top.
Free, on your device, no watermark
Because everything runs locally there is no upload, no account, no queue and no watermark. The same subject-detection model also powers the background remover, so if you've used that, the model is already cached and the effect is instant. Photos with one clear subject — a person, a pet, a product — work best.
How it works
Drop a photo — subject detection starts automatically (a ~5 MB model, downloaded once).
Type your text, then drag it on the canvas to position it behind the subject.
Tune size, font, colour and opacity, hit apply, and download in WebP, JPEG, PNG or AVIF.
Like every pixelpix tool, this runs 100% in your browser: your file is never uploaded to a server, never leaves your device, and works even offline once the page has loaded.
Frequently asked questions
Is the text-behind-image effect really free?
Yes — no watermark, no signup, no export limit. The AI model runs in your own browser, so there's no server cost to pass on to you.
Why does the text show in front of part of my photo?
The AI decides what counts as the subject. It's strongest with one clear subject against a distinguishable background; busy scenes with several overlapping subjects can confuse it. Try repositioning the text, or crop closer to the subject first.
Is my photo uploaded to a server?
No. Subject detection and the final compositing both happen in your browser with WebAssembly. Your photo never leaves your device.