background-remover

To remove backgrounds and make transparent cutouts, the Skill processes images using rembg models to generate alpha-ready outputs. It handles single-subject isolation and produces composite-ready PNG files for product imagery and marketing visuals.

Yes, batch processing for background removal is supported via a deterministic CLI in the scripts directory. You can process multiple images simultaneously to produce alpha-ready cutouts or finished JPEG/PNG composites.

Yes, background removal includes hair-edge alpha matting and color-spill removal. The pipeline applies pre-matting enhancements and edge refinement to ensure crisp cutouts without residual background color bleeding.

To crop a headshot from a group photo, the Skill uses face-aware cropping with OpenCV YuNet and a Haar cascade fallback. This isolates one person automatically while maintaining correct EXIF orientation.

Yes, after removing the original background, you can composite the cutout onto optional gradient backgrounds with shadows. The pipeline outputs finished JPEG or PNG files rather than just alpha channels.

Background removal supports standard image formats processed via Pillow and Pillow-HEIF, outputting alpha channels or full JPEG and PNG composites. OpenCV and NumPy handle the underlying image processing and matrix operations.