Run instantly with Docker
docker run -p 80:80 withoutbg/app:latestPress the copy button to copy this command to your clipboard
Background Removal API and Open-Weight Model
Two models, one kit. Self-host the open-weight model (Apache-2.0) for full data control, or use the managed API for stronger results on hair, fur, and complex edges.
Two models
/api-model · /open-weights-model
Precision cutouts
Hair, Fur, Glass, Motion
Run your way
self-host · API · plugins
Explore
/docs · /tech · /compare
Background Removal API and Open-Weight Model
Two models, one kit. Self-host the open-weight model (Apache-2.0) for full data control, or use the managed API for stronger results on hair, fur, and complex edges.
Two models
/api-model · /open-weights-model
Precision cutouts
Hair, Fur, Glass, Motion
Run your way
self-host · API · plugins
Explore
/docs · /tech · /compare
Open-weight model quick start
Python package
pip install withoutbgThe kit
Pick a model, then choose how you run it: demo, API, Docker, Python, or plugins.
API Model
Managed REST API hosted in Frankfurt, DE. p95 ≈ 800 ms server-side on AWS Inferentia. Strong on hair, fur, and fine edges. Pay per image, no subscription.
Inferentia · p95 ≈ 800 ms · EU (Frankfurt)
Open-Weight Model
Self-hosted background removal with full model weights released under Apache-2.0. Run locally via Python, Docker, or desktop plugins. Images stay on your network.
pip install withoutbgPress the copy button to copy this command to your clipboard
docker run -p 80:80 withoutbg/app:latestPress the copy button to copy this command to your clipboard
Specs and limitations
Specifications, security, training data, and known limitations for the API and open-source models.
Background removal models
Security and privacy
OSS: Run locally; no API calls required. Full control over image data.
API: TLS in transit; in-memory processing only; no disk writes of customer images.
Logging: Timestamp, endpoint, duration, status, request size, API key hash. No image bytes or perceptual hashes.
Analytics: No cookies on the web UI. Ahrefs (privacy-friendly). In-house CAPTCHA for abuse prevention.
Training data
No customer training: Uploads are never used to train models. Buffers zeroed after response.
Public photos: Unsplash/Pexels under permissive licenses; we created alpha mattes. withoutBG100 dataset
Licensed sets: Purchased image sets with explicit derivative rights.
Synthetic renders: Randomized lighting/camera with ground-truth mattes from the render pipeline.
Studio captures: Hair, translucency, shadows. Dataset guide
Scale: ~60K image/matte pairs after QA (2025-10-01), expanding.
Limitations
Transparent materials: Glass, water, and sheer fabric remain inconsistent due to complex light transport.
Subjective boundaries: When foreground vs. background is ambiguous, multiple valid cutouts exist. Future release will bias toward nearest-camera subjects.
FAQ
Is it really free?
Yes. OSS models are Apache-2.0. The hosted API is freemium: 50 free credits, then pay-as-you-go.
- Open source: Commercial use, modification, redistribution permitted.
- Hosted API: Newer weights, faster inference, autoscaling.
- Python package: Local OSS inference, or add
api_keyfor cloud. - Pricing: Credits per image. No subscription lock-in.
How fast is it end-to-end?
End-to-end depends on upload size and network RTT. We publish server-side processing only: p95 ≈ 800 ms in Frankfurt, DE. Total time = upload + TLS + queue + ~800 ms + download.
Do you store or train on my images?
No. Images are processed in RAM and discarded after the response.
- Storage: No disks, thumbnails, caches, or backups.
- Region: Frankfurt, Germany (EU).
- Logs: Metadata only (timestamp, endpoint, duration, status, size, key hash).
- Training: Customer uploads are never used for model training.
Do you use customer data for training?
No. Customer uploads are never used to train or fine-tune models.
Supported by
