Remove a video background on-device, free
One switch cuts your subject out of a clip — no green screen, no upload. voicedraw runs AI subject segmentation right on your device and feeds the matte straight into the render, so you can drop your subject onto a new background and export it transparent.
One switch, no green screen, no upload
Turn on Auto-Segment for an image or video layer and voicedraw separates the subject from its background using an AI segmentation model that runs entirely on your device. There is no green screen to light, no matte to paint, and — because it never leaves your browser — nothing to upload and no waiting on a server.
The cutout feeds a soft matte directly into the compositor, so you can immediately place your subject over a new background, a color, or an animated scene, and keep editing with the full toolkit around it.
On-device & private
Segmentation runs in your browser. Your footage stays on your machine — private and fast.
No green screen
Cut out a subject shot anywhere. Shot on green? There’s a dedicated Chroma Key with spill suppressor too.
Export transparent
Render WebM with alpha or a PNG sequence so your subject drops cleanly onto any new background.
Honest about what it does
This is subject segmentation, not frame-by-frame rotoscoping. It shines on clear foreground subjects — a person to camera, a product against a contrasting background — and produces a clean soft matte rather than hand-traced edges. For green-screen footage, the built-in Chroma Key and Spill Suppressor give you a tighter, more controllable key.
Common questions
How does the video background remover work?
Flip on Auto-Segment for an image or video layer and voicedraw runs an AI subject-segmentation model on your own device to separate the subject from its background, feeding a soft matte straight into the render. There is no green screen to set up and nothing to upload.
Is my footage uploaded anywhere?
No. The segmentation runs on-device in your browser, so your footage stays on your machine. That makes it private and fast, with no server round-trip.
What works best, and what are the limits?
It is subject segmentation, so it works best on clear foreground subjects — a person to camera, a product against a contrasting background. It produces a soft matte rather than frame-by-frame rotoscoped edges. If you shot on a green screen, voicedraw also has a dedicated Chroma Key effect with a spill suppressor for a cleaner key.
How do I export with a transparent background?
Once the subject is cut out, export WebM with a real alpha channel or a PNG sequence, both of which carry true transparency so your subject drops onto any new background.