Open a short video of a person and the AI (MediaPipe selfie segmentation) finds them in every frame. Make the background transparent, or replace it with a solid color, your own image or a blurred version of the original, then save a video or a ZIP of transparent PNG frames. Shot on a green screen? Use the color key instead. Everything runs in your browser.
Last updated:
0:00 / 0:00
How to find the background
Tip: click the screen color in the preview to pick it exactly.
New background
No image yet: a soft gradient is used.
Export
Your video
Advertisement
Advertisement
How to use the Video Background Remover
Choose a video with one or more people in it. Short clips (under 2 minutes) process fastest.
Press Remove background. The AI model downloads once, then each frame is analyzed; you can watch the cutout appear in the preview.
Pick a new background: Transparent, Color, Image (upload any photo) or Blur, and adjust Edge softness and Cutout tightness until the edges look clean.
Play the preview to check the result, then press Create video (it records in real time) or Download PNG frames (ZIP) for guaranteed transparency.
Green screen footage? Switch to Green screen key, click the screen in the preview to pick its color, and set the tolerance.
Choosing a background and export
Goal
Background
Export
Put yourself over other footage in a video editor
Transparent
PNG frames (ZIP), or transparent WebM if your editor reads VP9 alpha
Make your own green screen clip for a mobile editor
Color: green (#00B140)
MP4, then use the editor's chroma key
Clean, branded look for a presentation or social post
Color or Image
MP4 (or WebM)
Hide a messy room but keep a natural look
Blur
MP4 (or WebM)
How the cutout is made
Segmentation: each analyzed frame is scaled to about 512 pixels and passed to the MediaPipe selfie segmentation model, which returns a person-probability mask (general model: 256×256; fast model: 144×256 landscape).
Smoothing: each mask is blended with the previous one (65 % new, 35 % previous) to calm flicker between frames.
Edge shaping: Cutout tightness sets the probability where a pixel counts as person; Edge softness feathers the mask so hair and shoulders blend naturally.
Compositing: the original frame is kept where the mask is opaque and the new background is drawn underneath, at the original resolution (up to 1920 pixels on the long side).
Tips for better results
Film with light on the person and a background that is darker or lighter than their hair and clothes.
Avoid moving background people: the AI will keep any person it sees.
Trim long recordings to the part you need first; analysis time grows with every frame.
Arms and hands moving fast can blur; analyzing 30 frames per second follows them more closely.
Frequently asked questions
How does the AI remove a video background?
The tool uses MediaPipe Selfie Segmentation, a small neural network from Google that runs in the browser with WebAssembly. For each analyzed frame it outputs a confidence mask of where a person is. Masks are smoothed between frames to reduce flicker, scaled up with soft edges, and used to keep the person while the rest is made transparent or replaced.
What kind of videos work best?
Videos of people, especially one or two people facing the camera from the waist up, like webcam, vlog, talking-head or dance clips. The model is trained to find people, so it won't cut out pets, products or cars. Good lighting and a background that contrasts with hair and clothing give cleaner edges. For objects, film in front of a green screen and use the color key.
Can I get a video with a transparent background?
Yes, two ways. Download PNG frames (ZIP) gives an image sequence with a real alpha channel that Premiere Pro, DaVinci Resolve, After Effects, CapCut on desktop and most other editors can import. Transparent WebM keeps alpha when your browser's video recorder supports it (recent Chrome and Edge); always check the result, as other browsers may save a black background instead.
Why does exporting take as long as the video?
Browsers record video in real time with the MediaRecorder API, so the composited video plays through once while it is captured, together with the original sound. The AI analysis before that runs frame by frame and may be slower or faster than real time depending on your computer.
How do I fix flickering or rough edges?
Raise Edge softness for smoother, feathered edges, and move Cutout tightness up to remove background halos or down if hair and fingers get cut off. Analyzing 30 frames per second instead of 15 also reduces jumps in fast motion. Busy backgrounds and low light make edges less stable.
What is the difference between the AI and the green screen key?
The AI finds people anywhere, with any background, but edges around hair are approximate. The green screen key removes one color and is precise and instant, but only works when the subject was filmed in front of an evenly lit green or blue screen and doesn't wear that color.
Is my video uploaded?
No. The video is decoded, analyzed and recorded entirely in your browser. Only the AI model files are downloaded from a public CDN the first time; your footage never leaves your device and no watermark is added.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.