The viral design trick, automated: AI finds the subject of your photo and weaves your text behind them. Drag it into place, style it, download it. Runs entirely in your browser.
Drop a photo here
or
JPG, PNG, WebP · works best with one clear subject · the AI model (~4 MB) downloads once, then runs on your device
No photo handy? Try a sample imageYou have seen the effect everywhere: a huge word sitting behind a person in a photo, with their head and shoulders overlapping the letters. It reads as professional design because it used to require Photoshop layer masking. This tool automates the whole trick: an AI segmentation model finds the subject of your photo, and your text is drawn between the background and that subject, so the letters tuck naturally behind a head, a shoulder, a building, or a dog's ear.
Everything runs on your device. The segmentation model (about 4 MB) downloads to your browser once, and the subject detection plus all rendering happens locally through WebAssembly. Your photo is never uploaded, never queued on a server, and never logged.
Everything works without an account, without a watermark, and without your photo leaving your device.
Drop your photo in, and an AI model finds the subject automatically. Type your text, drag it anywhere, and it renders behind the person while staying in front of the background. Download the finished design as PNG or JPG with no watermark and no sign-up.
A segmentation model separates your photo into subject and background, right in your browser. The tool then draws three layers in order: the background, your text, and the cutout subject on top. The result is the same layering trick designers do manually in Photoshop, done automatically in seconds.
No. The AI model downloads to your browser once (about 4 MB) and the subject detection runs on your own device. Your photo is never uploaded, stored, or seen by anyone, and the tool keeps working offline once the page has loaded.
A clear main subject with some contrast against the background: a person, a pet, a product, a building against sky. Busy scenes with several equally prominent subjects give the AI less to lock onto. Position the text so part of it peeks out from behind the subject; that overlap is what sells the effect.
Yes. Add as many text layers as you want, each with its own font, size, color, opacity, rotation, and letter spacing. Click any text on the preview to select it and drag it into place.