Video Object Masks - Track objects and generate mask output
Video Object Masks
Track objects across a video and return mask data for downstream editing.
Upload MP4, MOV, or WebM. Duration-priced tools require uploaded files so the server can read video metadata.
Credits are calculated from server-read video metadata. Max 600 seconds.
Estimated credits
10 credits per 5 seconds · 120 credits per minute
Please sign in to use this tool.
No mask output yet
Upload a video and describe the objects to track.
Object mask output
Track objects through video and return structured mask data.
Video Object Masks is for workflows that need more than a caption. Describe the object you want to track and the tool returns mask-oriented output that can support editing, segmentation, review, or downstream automation.
Features
Prompt-based targets
Describe the object or region to follow, such as a person, vehicle, product, or item in the frame.
Mask-oriented output
Review structured outputs and raw payloads for downstream video editing or automation.
Longer clip support
Designed for object mask jobs up to ten minutes, with credits scaling by seconds.
Workflow
Add a video
Upload a video file so its duration can be read automatically for credit calculation.
Describe targets
Write the objects or regions you want the tool to track.
Inspect output
Review generated mask data, URLs, and raw payloads from the result panel.
Use cases
Use video object masks for visual workflows that need tracked regions instead of plain captions.
FAQ
Name the object or region clearly, such as person in red shirt, white car, logo, product box, or foreground subject.
This tool is focused on object mask-style output. The result panel also shows text or raw JSON when the processor returns it.
The current object mask page supports videos up to 600 seconds.
Credits are estimated at 2 credits per second with a minimum of 10 credits.
Track objects in a video.
Describe the target object and generate mask-oriented output for your clip.
Generate object masks