Segment Mask

Track an object through a video from a text prompt (Meta SAM 3) and get back a black and white mask video: white where the object is, black everywhere else. The prompt defaults to `face`; separate several objects with commas (`face, hands`). Clips up to 1800 frames (30 seconds at 60 fps, 15 seconds at 120 fps) as a regular MP4 or MOV. One flat price per clip, whatever its length.

Try Segment Mask in the playground

Last updated View as MarkdownAgent setup

Segment Mask

Node type: process:video_segment_mask
Category: process

Description

Track an object through a video from a text prompt (Meta SAM 3) and get back a black and white mask video: white where the object is, black everywhere else. The prompt defaults to face; separate several objects with commas (face, hands). Clips up to 1800 frames (30 seconds at 60 fps, 15 seconds at 120 fps) as a regular MP4 or MOV. One flat price per clip, whatever its length.

Pricing

  • Cost: 1 credit per clip, whatever its length (up to 30 seconds and 1800 frames). A clip past that is refused before you are charged.

Canvas ports

These appear as port handles on the left side of the node.

ID Label Details
video_url Video VIDEO (required)
prompt Prompt TEXT · default: "face". What to mask. Defaults to face. Separate several objects with commas, e.g. face, hands or person.

These render as form fields in the right-side config panel when the node is selected.

ID Label Details
detection_threshold Detection Threshold NUMBER · 0.1…1 · default: 0.5. Lower finds more (and less precise) matches, higher only confident ones.

Outputs

ID Label Type
video Mask Video VIDEO

What comes out

A black and white video the same size and length as the clip: white where the prompt matched, black everywhere else. It is the mask itself, not your clip with a colored overlay. Track several things at once by separating them with commas (face, hands), or use person for the whole body. If a small or turned face is missed, lower Detection Threshold; if it grabs things it should not, raise it.

What it is not, yet

It does not hide anything on its own. Laying the mask over the clip with Video Blend changes the colors of the whole frame (see the note on Video Blend), so it is not a clean face hide. A masked merge that keeps the rest of the frame untouched is not shipped yet.


Auto-generated from the Wireflow node registry.

For AI agents: the full documentation index is at /llms.txt, and most docs pages are available as Markdown by adding .md to the URL.

© 2026 Wireflow. All rights reserved.