@jackgreen2018/browser-background-remover

Browser-based background removal using pre-built WASM. No server uploads, no subscriptions, privacy by design.

Built for Etsy and Shopify sellers who process 10–50 product photos monthly and resent Remove.bg's $7.99/mo subscription.

Install

npm install @jackgreen2018/browser-background-remover

Usage

import { removeBackground } from '@jackgreen2018/browser-background-remover';

// From raw bytes
const result = await removeBackground(arrayBuffer, {
  threshold: 0.5,   // 0.0–1.0, lower = more conservative
  blurRadius: 2,    // Gaussian blur for soft edges
});

// result.imageData: Uint8Array of PNG bytes with transparent background
// result.alphaMap: Uint8Array of alpha values [0–255] per pixel
// result.width, result.height: output dimensions

// From a File object (browser)
const file = document.querySelector('input[type=file]').files[0];
const result2 = await removeBackgroundFromFile(file, { threshold: 0.4 });

API

removeBackground(input, options?)

Remove background from an image.

• input: ArrayBuffer | Blob — the source image (JPEG or PNG)
• options.threshold: 0.0–1.0 (default: 0.5) — sensitivity for background detection
• options.blurRadius: number (default: 2) — Gaussian blur radius for edge softening
• returns: Promise<BackgroundRemovalResult> with imageData, alphaMap, width, height

removeBackgroundFromFile(file, options?)

Convenience wrapper for browser File objects.

Benchmark

Measured on Node.js 22, 5 iterations each, real numbers:

| Image Size | Avg (ms) | Min (ms) | Max (ms) | |---------------|----------|----------|----------| | 100x100px | 4.55 | 2.95 | 10.70 | | 500x500px | 65.39 | 62.88 | 67.75 | | 800x600px | 121.72 | 117.61 | 135.64 | | 1000x1000px | 249.47 | 244.03 | 257.12 |

Comparison: Remove.bg API takes ~2–5 seconds per image (server round-trip + processing). This library runs entirely in-browser at ~5–250ms depending on image size.

How It Works

1. WASM decode — Pre-built WebAssembly module decodes JPEG/PNG input
2. Background estimation — Samples corner pixels to estimate background color
3. Alpha matte — Builds a soft matte via color-distance thresholding
4. Edge softening — Applies Gaussian blur to smooth transitions
5. PNG encode — Outputs transparent PNG

The WASM binary is pre-built and included in the package. No compilation required.

License

MIT