Remove Background

Make a plain or single-color background transparent, for example the white behind a product photo, a logo or a scanned signature, and download the result as a PNG. This tool works by color, not AI, so it will not neatly cut a person out of a busy photo.

Drop an image here or click to choosePNG, JPG, WebP, GIF or BMP

Your image is processed on a canvas inside your browser. It is never uploaded to our server.

How to use the Remove Background

  1. Drop an image onto the box or click to choose one. Photos of objects on a plain backdrop, logos, screenshots and scanned signatures work best.
  2. Look at the preview. The checkerboard pattern shows where the image is now transparent.
  3. If the wrong color was removed, click on the background in the preview to pick it yourself. Clicking also tells the tool to clear the area around that spot.
  4. Move the Tolerance slider until the background is gone but your subject is intact, and choose whether to remove only the outer background or every matching pixel.
  5. Press Download PNG. The file keeps its transparency, ready to place on a website, slide or document.

How it works

This is a color-based tool, the same idea as the "magic wand" in an image editor. It does not use AI or any kind of subject recognition. First it looks at the pixels along the four edges of the picture and takes the most common color there as the background. Every pixel is then compared with that color, and the ones close enough (as set by the tolerance slider) are made transparent.

In the default mode, only matching pixels that connect to the edge of the picture, or to the spot you clicked, are removed. That protects white details inside your subject, such as a white logo on a red mug. Switch to Every pixel of that color when you want enclosed areas cleared as well, for example the holes in letters like "o" and "e" on a logo, or the paper showing between the strokes of a signature.

Soften edges makes the outermost ring of pixels around your subject partly transparent when its color is close to the background. That hides the hard, jagged outline you would otherwise get, especially on JPEG photos.

When it works and when it doesn't

It does a good job when the background is one fairly even color that clearly differs from the subject: product shots on white or a colored sweep, logos and icons, charts and screenshots, a signature photographed on white paper. Even lighting helps a lot; a strong shadow or a gradient from light to dark will need a higher tolerance or will leave a patch behind.

It will not cleanly cut out people, pets, hair or anything in front of a busy, real-world background, because there is no single background color to remove. For those images you need an AI-based cut-out tool or careful manual masking in an image editor. Also expect trouble when the subject shares the background color, such as a white shirt on a white wall.

Frequently asked questions

Does this use AI?

No. It removes pixels that are close to a chosen background color. That is quick and predictable on plain backgrounds, but it cannot recognize a person or object, so it is not suitable for portraits or cluttered photos.

Why are parts of my subject disappearing?

Those areas are too close in color to the background. Lower the tolerance, or switch to Only background connected to the edges so that areas enclosed by your subject are left alone.

Why is there still a halo or patches of background?

The background is probably uneven, with shadows, noise or JPEG artifacts. Raise the tolerance a little, keep Soften edges on, or click directly on a leftover patch to remove it.

What format is the download?

A PNG file, because PNG supports transparency. JPEG does not, so a transparent background saved as JPEG would turn white or black again.

Is my image uploaded?

No. The image is drawn onto a canvas in your browser and processed there. Nothing is sent to our server.

Is there a size limit?

Images larger than about 16 megapixels are scaled down before processing so they fit in browser memory, and the tool tells you when that happens. Smaller images keep their original size.

Related tools