Quick answer: you can remove a background free in the browser without uploading the image. Local cutout runs an ONNX segmentation model on the device, produces a mask, and applies it to the original pixels, which is fast and keeps the file private.

How the in-browser pipeline works

  1. Load the model locally. A segmentation model in ONNX format is downloaded once and runs in the browser.
  2. Segment the subject. The model predicts which pixels belong to the foreground and builds a mask.
  3. Apply the mask. The background is removed from the original image, leaving a transparent cutout.
  4. Refine the edges. The editor lets you touch up the mask where hair or fine edges are tricky.

Why it differs from a remote service

  • No upload. The image never leaves the device, which matters for private or unreleased products.
  • Speed. There is no round trip to a server, so the cutout returns quickly.
  • No per-image credit. Processing on the device avoids a charge for each image.
  • Cost model. The limit is usually usage on the account, not a per-cutout fee.

The honest tradeoff

Remote services still lead on some complex edges, such as fine hair or transparent objects, because they can run larger models. A local model is very strong on clean product shots, which is the common ecommerce case, but difficult edges may need manual refinement.

What to do after the cutout

Place the subject on a new background, add the logo, and build the platform image. Keeping the cutout and the editor in the same tool avoids re-uploading the file. Neatico runs the cutout in the browser and finishes the image in a layer-based editor.

A quick workflow

  1. Open the image in the browser editor.
  2. Run the local cutout.
  3. Inspect the edges and refine if needed.
  4. Add the background and branding.
  5. Export at the platform size.

The Photoroom alternatives hub covers the local versus remote options.

The bottom line

In-browser background removal is a real, free-capable workflow that runs on the device and keeps the image private. It is the right default for clean product shots, with manual refinement for the rare hard edge.

What in-browser removal actually changes

Running the background-removal model in the browser with ONNX and a Web Worker means the cutout is local rather than a server upload, which helps with large product files and keeps the workflow on one page. The first run downloads the model, so it takes longer than later runs. The strongest results still come from a clean, well-lit subject; busy edges and transparent materials need a manual review and a quick touch-up before export.