Advertisement

← How It WorksAI Tool

How Video Background Remover Works

Flicker-free video matting in your browser — a recurrent People model or per-frame AI for any subject, exported as genuinely transparent WebM or composited MP4.

Try it free →

Step-by-step process

  1. 1

    Pick the model for your subject

    People clips use Robust Video Matting — a recurrent network that carries memory between frames, so edges stay stable instead of flickering (the structural fix, not a post-blur). Products, pets, and objects use per-frame segmentation: ORMBG for speed or BiRefNet for the sharpest edges, with temporal smoothing between frames.

  2. 2

    Frame-by-frame processing on your device

    Frames are decoded in order and processed at up to 1280px on the long side — the bound that finishes reliably without exhausting browser memory. Clips are capped at 60 seconds; trim longer footage first in the video editor. Nothing is uploaded at any point.

  3. 3

    Real-alpha export with audio

    Transparent output is encoded by ffmpeg.wasm as VP9 WebM with a real alpha plane (yuva420p) — not a baked-in checkerboard — and drops straight into CapCut, Premiere, Resolve, or our video editor. Composited output writes an H.264 MP4 with your solid, blurred, or image background. Original audio is kept either way.

100% private by design. All processing runs in your browser — your files never leave your device. No account required.

Frequently asked questions

Why does the People model not flicker?

Robust Video Matting is recurrent: each frame's result feeds into the next, so the model remembers where the subject was. Per-frame models decide each frame independently, which is where edge flicker comes from — we smooth it, but recurrence prevents it.

Can I use the transparent WebM in Safari?

Safari cannot decode VP9 alpha, so the transparent WebM shows an opaque background there. Export the composited MP4 for Safari viewers — the transparent file still works in Chrome, Edge, and every major desktop editor.

Why is there a 60-second limit?

Every frame runs AI inference and is re-encoded on your device. Sixty seconds at up to 1280px is what completes reliably across devices. Trim first, process the section you need.

How do I put my subject on a new scene?

Choose a solid or blurred background here for a quick composite, or open the transparent result in the video editor and pick any lifestyle scene — it shows through everywhere the clip is transparent.

Related tools