fal has added a new Pixelcut model aimed at removing backgrounds from video clips, expanding its AI tooling for creators and developers working with moving images.
The model, listed as Pixelcut Video Background Removal, is described as a segmentation system that processes video frame by frame while maintaining temporal consistency across the clip. In practical terms, that means the tool is designed to keep edges and subject separation steady from one frame to the next rather than creating flicker or other visible artifacts.
According to the product listing, the system is built to handle details that often challenge background removal workflows, including hair and movement. Those features suggest the model is intended for videos with more complex motion and fine subject contours, not just static shots.
The new offering appears on fal as a partner model and is available for inference and commercial use. Users can access it through both a playground interface and an API, giving developers a way to test the model manually or integrate it into their own applications.
fal's listing shows that the model accepts a video URL as input. The interface also supports common video file formats, including MP4, MOV, WebM, M4V and GIF. Additional settings are available for users who want more control over the input.
The sample output shown on the product page returns a WebM file, indicating that the tool produces a processed video rather than a still image or mask. The page also includes a download link for a demo result.
fal says the request cost is $0.022 per 30 frames. That frame-based pricing structure aligns with the model's video processing approach and provides a straightforward way to estimate costs for shorter or longer clips.
The launch reflects growing demand for automated video editing tools that can reduce manual work in post-production. Background removal has long been a common feature in image editing, but extending the same capability to video is more technically demanding because the system has to preserve consistency over time while tracking movement, changing light and fine details around the subject.
By making the model available through its API, fal is positioning the capability for developers building editing software, social media tools, virtual production workflows and other applications that rely on fast video manipulation. The commercial-use designation also suggests that the company expects the model to be used in production environments rather than only for experimentation.
Pixelcut's model is part of a broader trend in generative and editing tools that are bringing more granular control to video workflows. As AI systems improve at separating subjects from backgrounds, they are increasingly being used to support tasks that once required labor-intensive manual rotoscoping or specialized visual effects software.
For now, fal's release gives users another option in the video editing toolkit, with a focus on one of the harder parts of the process: preserving clean subject cutouts across motion-heavy footage.