Extract Object, Bria's prompt-based object extraction API

Bria launches Extract Object on Fal, a prompt-based API that isolates image elements into transparent PNGs to automate visual production pipelines.

Bria makes available on Fal an API named Extract Object, capable of isolating a specific element from an image using a simple natural language description. A prompt like "the yellow hammer" is enough: the designated object is extracted then returned as a PNG RGBA, with a transparent background, ready to be recomposed without manual editing.

The goal is the automation of visual production pipelines. Object-by-object cutout previously relied on a dedicated tool and a retoucher, taking five to thirty minutes per visual, a bottleneck whenever a catalog exceeds a thousand items per week. A single request now replaces this step.

The developer describes three uses: large-scale product isolation for catalog teams, the retrieval of elements from existing campaign visuals to create advertising variations, and the integration of cutout as a native building block within a pipeline. The API integrates with the company's Background Generation, Product Shots, and Relighting modules to cover an end-to-end workflow.

Commercially, Bria claims the top spot in product visual extraction, ahead of SAM 3.1 and other open-source solutions, according to its own metrics. As output, the canvas can be automatically cropped around the object, with mask refinement through background removal remaining optional.