Background Remover

How AI background removal works

Background removal has historically required Photoshop skills or cloud APIs that upload your image to a third-party server. This tool uses @imgly/background-removal, an open-source library that runs a pre-trained ONNX segmentation model directly in your browser via WebAssembly. No image data ever leaves your device.

What is ONNX?

ONNX (Open Neural Network Exchange) is an open format for machine learning models. This tool packages a segmentation model trained to identify foreground subjects (people, products, animals) and separate them from the background. The model runs inside a Web Worker so it doesn't block the browser UI during inference. Inference time depends on image resolution and your device — typically 3–15 seconds on a modern laptop.

What output do you get?

The output is always a PNG file with a transparent background (RGBA channels). The checkerboard pattern in the preview is the standard way to visualize transparency — it means those pixels are fully transparent in the output. You can place the downloaded PNG on any coloured background in Canva, Figma, PowerPoint, or any image editor.

Best results

• Clear subject against a contrasting background
• Good lighting with minimal motion blur
• Product photos on white or solid-colour backgrounds work especially well
• Complex hair and fine edges may have slightly rough edges — this is a known limitation of browser-side models

Privacy

Your image is processed entirely in your browser. The model is downloaded once and cached locally. Subsequent uses don't require re-downloading. Nothing is sent to any server at any point.