Text Behind the Subject in Photos and Video

An online tool for the text-behind-subject effect. A neural network separates the person or object from the background, and your text slides behind them — the same sense of depth you see on posters and album covers. It works on photos and video alike: on video the subject is tracked frame by frame, so the words stay behind a moving person.

How to do it

1
Upload a file

Drop a photo or video — the network separates the subject from the background.

2
Write your text

Pick a font, colour and size, then drag the words where you want them.

3
Check the depth

The text slides behind the subject; you can keep some layers in front instead.

4
Download it

Photos as PNG or JPG, video as MP4 with sound.

Place your words behind a person or object in a couple of clicks

Drop a photo or a video
AI separates the person or object from the background
JPG · PNG · WEBP · MP4 · WEBM · MOV ·

Photos take seconds. Video is matted frame by frame — slower, but the words stay behind the person as they move.

Loading the AI model…

Drag the text with a mouse or finger straight on the image.

Text layers
Styling
Colour
Presets
Export

Photo cutout uses the ISNet model via @imgly/background-removal.

Features

Photos and video in one place
ISNet handles stills, RobustVideoMatting handles video with frame-by-frame cutout, so the words stay behind the subject even as they move.
AI subject cutout
Nothing to trace by hand: the model finds the edges of a person or object, including hair and thin details.
Real typography
Ten fonts, gradient fills, outline, shadow, tilt, letter spacing and opacity — plus several text layers at once.
No tofu boxes for Cyrillic
The tool warns when the chosen font has no Cyrillic glyphs and substitutes a fallback, so you see real letters instead of empty squares.
One-tap style presets
Poster, outline, sunset gradient and soft — four ready looks you can then fine-tune.
Export at source quality
Photos save as PNG or JPG at the original resolution; video exports as MP4 (H.264) with the audio kept.

Questions and answers

How does the text end up behind the person?

The network builds a mask of the subject — which pixels belong to the person or object. The layers are then stacked in order: background, then your text, then the cut-out subject on top. The subject covers the words, so they read as being behind it.

Does it work with video?

Yes. The subject is matted frame by frame with RobustVideoMatting, so the words stay behind a moving person. It is slower than a photo: the longer and larger the clip, the longer it takes.

Why do some letters look like a different font?

Several display faces (Anton, Bebas Neue, Permanent Marker and others) are Latin-only and carry no Cyrillic glyphs. We substitute a fallback font so you never get empty boxes, and we warn you — pick a font with Cyrillic, such as Rubik Black or Russo One, to keep one style.

Can I keep the text in front instead?

Yes, every layer has a "Behind subject" switch. Turn it off and the words sit on top. You can mix the two: one word behind the person, another in front.

What if the cutout is inaccurate?

Mask quality depends on the contrast between subject and background. Shots where the person stands out clearly work best. If the edges bleed, try a different frame or a photo with a cleaner outline.

You May Also Like

✨ AI Photo Enhancer & Upscaler 🪪 Document photo 📸 HEIC to JPG Converter 🎨 Photo Colorizer 🩹 Photo Repair ©️ Photo Watermark 📷 Screenshot Tool 📱 Social Media Photo Resizer

How do you rate this tool?

Thank you for your rating!
Want to share more? Leave a comment!
Thank you! Your comment will appear after moderation.
Who is this tool for?
🎬 Video Creators 📡 Streamers & Broadcasters ✨ Content Creators
Published Updated Author: