Portrait isolation is one of the oldest tasks in digital image editing. It is also one of the hardest to get right. Commercial tools either charge a subscription for what should be a single-click operation, or they ship your image to a server you know nothing about.
SoloBG exists because neither of those trade-offs is acceptable. A person's photograph should not have to pass through a stranger's infrastructure to get a clean cutout.
The tool runs entirely in your browser using a portrait matting model downloaded once and cached locally. From that point forward, every image you process never leaves your device. There is no account, no upload, no log, and no charge.
SoloBG uses a state-of-the-art portrait matting model running via ONNX in a dedicated Web Worker thread. The model was designed specifically for human subjects — not generic objects — which is why it handles fine hair, glasses, and semi-transparent fabric far better than general-purpose background removal tools.
On first visit the model weights are downloaded from Hugging Face and stored in your browser's local cache. Subsequent visits load instantly. All image processing — decoding, inference, mask generation, alpha compositing — happens in a background thread on your own hardware, using WebGPU when available and falling back to WASM otherwise.
The output is a lossless PNG with a genuine transparency channel. No white fringe. No watermark. No compression artefacts added by a server roundtrip.
SoloBG is one of many tools built under the RuntimeHub umbrella — a growing collection of browser-first, privacy-respecting utilities. Every tool in the suite follows the same three principles: zero upload, zero account, zero cost.
The suite is designed and maintained by a single independent developer. If SoloBG is useful to you, the best way to support it is to share it with someone who needs it.