unbagrnd
What it is
unbagrnd removes photo backgrounds with on-device AI. Its desktop app processes images locally using ONNX Runtime; the maker says images, filenames, and metadata are not sent anywhere. The site says the only network request is a one-time model download, with no analytics, update checks, or telemetry. You can process a single image or a folder and see before-and-after previews as each file finishes. Several models are available to balance speed and accuracy, starting with a 43 MB default. Brush tools erase or restore subject areas, with undo and redo; you can also add a solid-color background and an on-device drop shadow. The app predicts a foreground mask on the CPU, composites it with the original pixels, and saves a PNG. The free, MIT-licensed product has no paid tiers or usage limits. It also offers a documented self-hosted REST API for single-image and batch processing. Available platforms include Android, Linux, macOS, Windows, and self-hosted use.
For and against
For
- Processes desktop images locally with ONNX Runtime
- Handles a single image or a folder
- Offers brush editing with undo and redo
- Free and open source under the MIT license
- Includes a self-hosted REST API
Against
- Android APK is not on Google Play
- macOS build is not notarized
- Windows installer is not code-signed
- API defaults to 20 MB per image and 10 per batch
The verdict
unbagrnd combines local image processing with manual edits and folder handling, without paid tiers or usage limits. Check the platform install notes and API defaults if those affect your setup.
Who it suits
It suits people who want local photo background removal, batch processing, or manual touch-ups. Teams needing an API can use the self-hosted REST API.
Plans and prices
unbagrnd
No paid tiers · no usage limits
Read 4 Oct 2026 · zidniryi.space
On the record
- API limits
- The documented API defaults to a 20 MB per-image upload limit, batches of up to 10 images, and 60 requests per minute per API key.zidniryi.space · read 4 Oct 2026
- API privacy
- The API docs say images are processed in memory and discarded, not written to disk or logged; SQLite stores API key metadata only.zidniryi.space · read 4 Oct 2026
- Editing tools
- Users can brush to erase or restore subject areas with undo and redo, then add a solid-color background and an on-device drop shadow.zidniryi.space · read 4 Oct 2026
- Install notes
- The maker says the macOS build is not notarized, the Windows installer is not code-signed, and the Android APK is not on Google Play.zidniryi.space · read 4 Oct 2026
- License
- The product is free and open source under the MIT license, with no API keys, paid tiers, or usage limits.zidniryi.space · read 4 Oct 2026
- Local processing
- The desktop app runs background removal locally with ONNX Runtime, and says images, filenames, and metadata are never sent anywhere.zidniryi.space · read 4 Oct 2026
- Model selection
- Settings lets users choose among several background-removal models to trade speed for accuracy, starting with a 43 MB default.zidniryi.space · read 4 Oct 2026
- Offline and telemetry
- The site says the only network request is a one-time model download and that the app has no analytics, update checks, or telemetry.zidniryi.space · read 4 Oct 2026
- Output and inference
- The app predicts a foreground mask on the CPU, composites it with the original pixels, and saves the result as PNG.zidniryi.space · read 4 Oct 2026
- Purpose
- unbagrnd removes photo backgrounds using on-device AI.zidniryi.space · read 4 Oct 2026
- Self-hosted API
- The maker documents a Docker-capable self-hosted REST API with single-image and batch background-removal endpoints.zidniryi.space · read 4 Oct 2026
- Single and batch processing
- The app processes one image or a folder and shows live before-and-after previews as files finish.zidniryi.space · read 4 Oct 2026
