Remove Background from Image

Last updated: October 1, 2026

Remove image backgrounds automatically using AI or manual color selection.

✅ Free Forever 🔒 Files Stay on Your Device ⚡ No Signup Required 🌐 Works in Any Browser

Drop your file here

or click to browse

Accepted: image/*

What is Remove Background from Image?

Remove Background cuts the subject out of a photo and gives you back a transparent PNG. There is nothing to draw or trace — upload a photo and the subject is detected automatically, then you adjust one slider if the edge needs tightening. The detection model runs on your own device, so the photo is not uploaded anywhere.

It is built for photos of people

The on-device model is a portrait segmentation model, which means it is genuinely good at people and does not recognise anything else. Headshots, full-body shots, group photos and selfies work well. A dog, a product on a table or a screenshot will find no subject at all — and when that happens the tool says so rather than quietly handing you a bad cutout.

In that case it falls back to colour detection, which removes background by colour similarity starting from the edges of the frame. That works when the background is reasonably even — a product against a plain white sheet, a logo on a flat colour — and struggles when it is busy. Knowing which of the two produced your result matters, so the tool tells you underneath the preview every time.

One slider, and what it actually does

After an automatic cutout the slider is labelled Edge tightness and moves the cut point of the detected mask between 35% and 85%, starting at 60%. Raise it to bite further in and clear background colour still clinging to the outline; lower it to keep more of the soft edges, which matters for hair. It refines the detection — it never switches it off. When colour detection was used instead, the same slider becomes Colour tolerance and controls how far the fill spreads.

You can see the result before you download

The original and the cutout sit side by side, with the transparent areas shown over a checkerboard so you can actually see what has been removed rather than guessing against a white page. The download is a PNG, because transparency cannot survive in a JPG.

What it will not do perfectly

Fine hair against a busy background is hard for any browser-based tool, and this one is no exception — the paid services that handle it well run much larger models on their own servers. Expect a clean cutout on a clear subject and expect to nudge the slider on a difficult one.

Related tools

If you want to put something behind the subject rather than nothing, Change Image Background does the cutout and the replacement in one step. Blur Image Background keeps the original background but throws it out of focus, and Profile Photo Maker crops a finished cutout to the square and circular sizes profile pictures need.

How to Use Remove Background from Image

  1. Upload Image

    Select an image with a background to remove.

  2. Remove Background

    Click the background color or use auto-remove. Adjust tolerance for accuracy.

  3. Download

    Download as PNG to preserve transparent background.

Features

Background Removal

Click to select background color for removal.

Tolerance Control

Adjust how similar colors are included in the removal.

Transparent PNG Output

Download with transparent background as PNG.

Frequently Asked Questions

Is my photo uploaded to remove the background?

The detection model runs on your own device, so in normal use the photo is not uploaded. The model itself is downloaded to your browser the first time you use the tool, which is why the first run takes a few seconds longer than later ones.

Why did it not detect my dog, my product or my logo?

The automatic detection is a portrait model - it recognises people and nothing else. For any other subject the tool says no person was found and switches to colour detection, which removes the background by colour instead and works best when that background is fairly even.

What does the Edge tightness slider do?

It moves the cut point of the detected mask between 35% and 85%, starting at 60%. Raising it tightens the outline and clears leftover background colour; lowering it keeps more of the subject's soft edges, which helps with hair. It refines the automatic detection rather than replacing it.

What file do I get back?

A PNG with a genuinely transparent background. PNG is the only option here because JPG cannot store transparency - a JPG would fill the removed area with solid white.

How good is it on hair and fine detail?

Good on a clear subject against a reasonably plain background, imperfect on fine hair against a busy one. That is a real limit of running the cutout on your own device rather than on a server, and the edge slider is there to help you tune the difficult cases.

Do I need an account?

No. There is no signup, no watermark on the result and no limit on how many photos you can cut out.