Always Free · No Signup

Click anything. Cut it out.

Point at the object you want and its outline is traced for you. Add more clicks to include or exclude parts until the selection is exactly what you meant.

Runs entirely in your browser. Your files never leave your device.

Point at one object — not the whole subject. Pick a single person out of a group, or one item off a shelf.

The first photo takes a little longer while the tool gets ready. It is remembered for next time.

Each photo is analysed once when you upload it. After that, every click returns a selection almost instantly.

How to cut out an object from a photo

  1. 1

    Upload a photo

    The photo is analysed once in your browser. This is the only slow step, and it never happens again for the same image.

  2. 2

    Click the object you want

    One click on the object is usually enough. Add clicks to pull in a part that was missed, or switch to Exclude and click anything that should not be there. The selection updates as you go.

  3. 3

    Download the cutout

    The selection is exported as a transparent PNG, cropped to the object so you can drop it straight into a design or another photo.

Object selection questions

The background remover decides what the subject is and removes everything else — one click, no control. This tool does the opposite: you point at the specific thing you want. In a group photo the background remover keeps all four people, while here you click one of them and get only that person.

Add more clicks. Another Include click on a missed area extends the selection; switching to Exclude and clicking an area removes it. Selections often need two or three clicks — that is the tool working as intended, not a failure. Clicking a marker again removes that click.

Boundaries with low contrast are the hard case: a grey cat on a grey sofa, a white mug on a white table, or a dark jacket against a dark doorway. Thin structures like hair strands, wire fences, and chain links also come out approximate. Objects seen through glass or reflections are unreliable. Where the edge is genuinely ambiguous, the result is a guess.

The whole photo has to be read once before any click can be answered, which takes a few seconds on a fast device and 10 seconds or more on a slow one. That work is kept, so every click afterwards comes back in well under a second.

No. The cut-out is drawn onto a new canvas and encoded from scratch, so camera model, timestamp, and GPS coordinates are all left behind. That is usually what you want when a cut-out is heading into a public design, though it does mean you should keep the original if you need the capture date for your own records.

Two things happen on export. The image is capped at 2048 pixels on the long edge before analysis, and the PNG is cropped to the bounding box of what you selected — so a cutout of one bottle from a shelf photo is bottle-sized, not shelf-sized.

Clicks combine into a single selection, so clicking two separate objects gives you one PNG containing both. If you want them as separate files, cut out one, click Clear all, and select the next.

More free tools

Beyond the Mask

Fill the Selection With Something New

A mask marks the region; it does not fill it. The studio takes the same photo and rewrites the part you point at — swap the product, change a garment's colour, replace the sky — from a written instruction.

  • Replace a selected object by describing the new one
  • Change colours and materials with a written instruction
  • Render the whole scene again with different content
Explore Studio Features
The same portrait before and after, with clothing and setting replaced from a written instruction