What is Background removal?
Also known as: Subject extraction, Alpha matting
An AI technique that separates a subject from its background, producing an alpha-channel cutout. Used in product photography, motif extraction, and design compositing.
In detail
Background removal models (BRIA RMBG, Modnet, U2-Net, Carvekit) take an input image and output an alpha mask separating subject from background. Modern models achieve hair-level edge fidelity — wisps, fringe, lace, and embroidery threads are correctly extracted with soft alpha edges. Textile use cases include: lifting a fabric swatch off a studio background for product listings, extracting a motif from a hand-drawn sketch for vectorization, isolating a model wearing a printed garment for a clean catalog shot, and cleaning user-uploaded reference images before AI generation. BRIA RMBG 2.0 is the current state-of-the-art for commercial textile work.
Example
A product photo of a folded silk scarf on a marble countertop. Background removal outputs the scarf with alpha=255, the marble with alpha=0, and soft alpha (50-200) around the scarf's tasseled fringe — preserving every thread. The cutout drops cleanly onto an e-commerce product page background.