Skip to main content
ModelRunner
bria avatar

bria / video/background-removal

Automatically remove or replace the background in a video — no green screen or manual rotoscoping — with output on transparent alpha or a solid color.

remove background
0.0042 per second of output video
OpenAPI

Model Input

Input

Input video to remove the background from. Must be under 4000x4000 resolution and shorter than 30 seconds.

If true, the audio track from the input video is preserved in the output.

Additional Settings

Customize your input with more control.

Fill behind the subject after the background is removed. Choose Transparent for an alpha-channel cutout, or a solid color to flatten the background.

Container and video codec of the output. Only alpha-capable containers (webm_vp9, mov_proresks) carry a transparent background.

You need to be logged in to run this model and view results.
Log in

Model Output

Output

Loading
Generated in 18.717 seconds
Logs (1 lines)

Model Example Requests

Examples

Example output 1Example output 2

Model Details

Model Details

Bria Video Background Removal (VRMBG 3.0) strips the background from a video automatically, frame by frame, with no green screen and no manual rotoscoping. Supply a video URL and the model returns the same clip with the background removed — either as a transparent alpha channel for compositing, or filled with a solid color you choose. Audio from the source is preserved by default, and clip length is unchanged. It is trained entirely on licensed data, so the results are safe to use commercially and at scale.

## Best for - Cutting out a presenter, talking head, or podcast host so they can be composited into a new scene - Producing transparent-background (alpha) clips for editors, motion graphics, and overlays - Cleaning up product and e-commerce videos onto a solid, distraction-free background - Swapping a busy background for a single flat color without shooting against a green screen - Commercially-safe video cutouts where licensed training data matters

## Choose another model when - You need to remove the background from a still image rather than a video — use an image background-removal model - You want to lip-sync, re-voice, or dub a talking-head video rather than cut out its background — use a video lip-sync model - You need to generate a new video from a prompt or a still image — use a text-to-video or image-to-video model

## Tips - The input video must be under 4000×4000 resolution and shorter than 30 seconds. - Choose `background_color: "Transparent"` for an alpha-channel cutout to composite over other footage; pick any of the 10 solid colors to flatten the background instead. - The default output is `mp4_h264` (broadly compatible, opaque). Transparency requires an alpha-capable container — `webm_vp9` or `mov_proresks` — so pair `background_color: "Transparent"` with one of those; `mp4_h264`/`mp4_h265` and `gif` cannot carry an alpha channel. - Set `preserve_audio: false` to drop the original audio track from the output.

## Advanced Configuration - `background_color` — the fill behind the subject once the original background is removed. `Transparent` yields an alpha cutout; the other values (`Black`, `White`, `Gray`, `Red`, `Green`, `Blue`, `Yellow`, `Cyan`, `Magenta`, `Orange`) flatten to that solid color. - `output_container_and_codec` — the container and video codec of the result. Options: `mp4_h265`, `mp4_h264` (default), `webm_vp9`, `mov_h265`, `mov_proresks`, `mkv_h265`, `mkv_h264`, `mkv_vp9`, `gif`. Only alpha-capable containers (`webm_vp9`, `mov_proresks`) carry a transparent background.

```js import { modelrunner } from "@modelrunner/client";

const result = await modelrunner.subscribe("bria/video/background-removal", { input: { video_url: "https://media.modelrunner.ai/00eWLaMhCPsIt1hB-woman-yellow-raincoat-walking.mp4", background_color: "Transparent", output_container_and_codec: "webm_vp9", preserve_audio: true, }, }); ```