Put Text Behind a Person in a Video
Upload a clip and the subject is cut out on our GPUs, frame by frame, with one AI model tracking it the whole way through — so the edges never flicker. Add text behind them, drag it into place, and export an MP4. Free, no watermark, up to {seconds} seconds.
Tool by Pixly — AI property staging, videos & reels in minutes.See howDrop a clip here
or click to browse — MP4, MOV or WebM, up to 15 seconds
Sign in to use video cutouts. Video runs on our GPUs, so it needs an account.
Upload a clip to start — the subject is tracked automatically.
Text layers
Text
Free for personal and commercial use. The clip is processed on Pixly's servers and deleted within minutes.
Your clip goes to our GPUs — and is deleted within minutes
This is the one Pixly tool that cannot run entirely on your device. Tracking a subject through hundreds of frames needs a GPU, so the clip is uploaded, processed on our own servers in Europe, and deleted as soon as the matte is made. The text you add never leaves your browser: only the cutout is done on our side, and nothing is kept.
Why the edges don't flicker
Most tools that do this cut out every frame separately. Masks computed independently disagree slightly from one frame to the next, so the edges crawl and shimmer — the artefact that makes an effect look cheap. We cut out the first frame properly, then a video model carries that one mask through the whole clip, which is stable by construction.
Frequently asked questions
Is it free?
Free accounts get three clips a day. A Pixly plan lifts that to a limit nobody reaches. There is no watermark either way.
Why do I need an account?
A clip is around 24 seconds of dedicated GPU time, against about two for a photo. That cannot be handed to anonymous traffic — the free tier would be farmed within a day. The image tool needs no account precisely because it runs in your browser.
How long can the clip be?
Up to 15 seconds. Cost is linear in frames, so length is the real limit rather than an arbitrary one: 15 seconds at 30 fps is 450 frames to track.
Does it work on things that aren't people?
Yes. The model tracks whatever the first frame's cutout selected — a person, a car, a house, a product. It works best when the subject is clearly visible in the opening frame, because that is what the rest of the clip inherits.
Can I put text in front as well as behind?
Yes. Each text layer has its own setting, so a headline can sit behind the subject while a caption sits in front of it. Add as many layers as you need.
What if the subject leaves the frame?
The model keeps a memory of the subject and picks it up again when it comes back. A subject that leaves and returns in a different pose is the hardest case; scrub through before exporting to check.
Free Image & Video Tools by Pixly
From Pixly
Turn one property into staged photos, videos & reels
Pixly creates AI virtual staging, day-to-night, construction reveals and cinematic property videos — in minutes.
BeforeAfter