Remove White Background from Image

Upload a picture shot on white and get the subject back on a transparent background. No sign-up, no watermark, and it usually takes a few seconds.

Sent securely, processed, and never saved.

Where white backgrounds come from

White backdrops are everywhere. Product photographers shoot on white sweeps, scanners turn paper into a white rectangle, and screenshots and brand graphics are often exported on white. Removing that white is usually the first step before putting the subject on a coloured banner, a mockup or a marketplace listing with a different look.

Why AI segmentation beats magic-wand selection

The classic way to remove white is a threshold: select every pixel close to pure white and delete it. That works on a flat logo, but it fails on real photos. Soft shadows are grey, not white, so they stay behind as a dirty halo. And anything white inside the subject, such as a shirt, a cup or a highlight in an eye, gets punched out along with the background.

The model used here, BiRefNet, runs on Cloudflare and decides what is subject and what is background from the shape and context of the whole image. A white sneaker on a white floor keeps its white. A shadow underneath is treated as part of the scene rather than a stray blob of grey.

  • White-on-white edges are handled by shape, not by colour distance.
  • Soft shadows are not mistaken for the subject.
  • Fine detail such as hair and fabric edges stays soft instead of jagged.
  • Images up to 20 MB and 4096 px on the long side keep their full size.

Tips for cleaner results on white

  1. Use the original file rather than a screenshot of it. Compression artifacts around edges make cutting harder.
  2. If the subject is mostly white, a light, even shot helps the model see the outline.
  3. Zoom in on the preview at the edges after processing. Subtle halos are easier to spot when enlarged.
  4. For a flat white graphic with crisp edges, the result is usually clean, but check the corners and thin lines.

Privacy and limits

Your image is sent over an encrypted connection to the site's server, the white background is removed, and the result returns straight to you. Images are not saved or kept, and location and camera metadata are removed from the copy that is sent. Each visitor gets a free daily allowance of 30 images.

When your goal is a PNG for a design, see remove background from PNG. When you are cutting the subject out to reuse it, the photo cutout page covers composing the result into collages and thumbnails.

How to remove a white background from an image

  1. Add the image

    Drop your studio shot, scan or screenshot into the upload box. JPG, PNG and WebP are accepted.

  2. Let the model work

    It looks at the whole picture, not just the colour of each pixel, and separates the subject from the white behind it.

  3. Check the edges

    Look at the preview against the checkerboard. White parts that belong to the subject should still be there.

  4. Export

    Download a PNG for transparency, or choose another preset first if you want a different backdrop.

Questions

How do I remove a white background from an image?

Upload the image, wait a few seconds for the AI model to separate the subject, and download the result as a transparent PNG. No account or watermark is involved.

Will it delete white parts of my subject too?

It is designed not to. The model separates subject from background by shape and context, not by colour alone, so a white shirt or product keeps its white. Check the edges in the preview to confirm.

Can I remove a white background from a scanned picture?

Yes, scans work like any other image. Use a clean, flat scan and zoom in on the edges after processing, since paper shadows and scan noise can leave faint marks.

What formats can I upload and export?

Upload JPG, PNG or WebP. Export PNG to keep transparency, or JPG and WebP if you prefer. JPG fills transparent areas with white.

Is it free?

Yes. It is free, with no sign-up and no watermark, and each visitor gets a daily allowance of 30 images.

Related