RMBG 2.0 Background Removal

State-of-the-art background removal by Bria AI, trained on licensed data for commercial-safe results

Generate
Result

Results will appear here

About This Model

RMBG 2.0 is Bria AI's state-of-the-art background removal model, built on the BiRefNet architecture and enhanced with Bria's proprietary licensed dataset. It consistently outperforms other models on complex backgrounds, especially fine details like hair and semi-transparent objects.

Key Features

  • Best-in-class accuracy — Outperforms BiRefNet, Adobe Photoshop, and other models in benchmarks
  • Commercial-safe — Trained exclusively on licensed data, safe for commercial use
  • One-click operation — Just upload an image, no parameters to configure
  • Transparent PNG output — Clean alpha channel ready for compositing

Best For

  • E-commerce product photos requiring the cleanest possible cutouts
  • Professional portrait extraction with complex hair edges
  • Batch processing via API where quality matters more than cost
  • Any use case where edge precision is critical

FAQ

How does RMBG 2.0 compare to BiRefNet?

RMBG 2.0 is built on top of BiRefNet but trained with Bria's proprietary licensed dataset. It handles complex backgrounds significantly better — especially hair, fur, and semi-transparent objects. The trade-off is cost: RMBG 2.0 is about 9x more expensive per image.

When should I use BiRefNet instead?

Use BiRefNet when cost matters more than edge perfection — it's ~9x cheaper and still very good for most images. BiRefNet also offers model variants (Portrait, Matting) and resolution control for fine-tuning.