withoutBG Open-Weight Model vs BiRefNet
23 examples. Use the view controls on each row to switch between transparency, chroma key, and alpha matte on all outputs.
Photo by Alexander Dummer on Unsplash
Photo by Andreas Rasmussen on Unsplash
Photo by Camila Quintero Franco on Unsplash
Photo by Carlo Sierra on Unsplash
Photo by Conor Samuel on Unsplash
Photo by Lars Bo Nielsen on Unsplash
Photo by Mayer Tawfik on Unsplash
Photo by Niko Tsviliov on Unsplash
Photo by Norbert Braun on Unsplash
Photo by Vinicius Wiesehofer on Unsplash
Methodology
- Alpha mattes are grayscale masks encoding per-pixel foreground probability. They capture fine edge detail (hair, fur, semi-transparent regions) that binary masks discard.
Photo Credits
Source images are courtesy of the photographers below via Unsplash.
BiRefNet run (published results)
- Code: ZhengPeng7/BiRefNet (local clone + infer_dir.py)
- Weights: ZhengPeng7/BiRefNet_dynamic via Hugging Face
- Hardware: Apple Silicon, PyTorch MPS
- Date: 2026-07-26
Inference settings
- Input size: 1024×1024 (square resize), then mask resized back to original resolution
- Normalization: ImageNet mean/std
- Precision: FP32 (no FP16)