Remove an unwanted object from a photo

The perfect shot almost always has something spoiling it: a random passer-by behind you, wires over a landscape, a bin, a watermark, an ex in a group photo. You used to go to a retoucher for this; now a neural network removes the extra in seconds — and, more importantly, fills in what was behind the objectas if it had never been there.

2–3 secto remove an object

1 phraseinstead of brush and masks

$0no retoucher

How the network "fills in the emptiness"

The technology is called inpainting. You mark the area to remove — the network erases it and fills it with a plausible continuation of the background: it extends the wall, the grass, the sky, repeats the pattern. It doesn't "find" what was hidden behind the object — that's not in the photo — but synthesizes a natural continuation from what's around. So the more uniform the background, the more seamless the result.

The same beach at sunset with no bin — the network filled the spot with clean sand — Drag the slider: the bin disappears, and the network builds sand in its place — as if it had never been there. The background is uniform, so there's no visible seam.

A beach at sunset with an ugly bin in the foreground — Drag the slider: the bin disappears, and the network builds sand in its place — as if it had never been there. The background is uniform, so there's no visible seam.

Upload a photo and in one sentence say what to remove — for example "remove the wires" or "delete the person on the left". The network removes the object and fills the background.

Загрузка…

When it's easy, and when it's hard

Easy

An object on a simple or repeating background — sky, grass, a wall, asphalt. The fill is almost indistinguishable.

Harder

The object covers something important — a face, text, a complex pattern. The network will imagine the continuation but may err on the details; sometimes it takes a few tries.

The limit

Removing a large object in the centre of a busy scene — the model will "invent" what it doesn't know, and that can look strange. Here it helps to remove in parts and clean up the result.

What people usually remove

Extra people — passers-by, tourists in the frame, an ex in a group photo.
Technical clutter — wires, poles, signs, bins, flash shadows.
Watermarks and logos — but carefully: removing a watermark from someone else's paid photo to use it for free violates rights. Do this only with your own images.
Defects and stains — alongside restoring an old photo.

Опрос

What would you most often want to remove from photos?

Проголосуйте, чтобы увидеть результаты

A little pro trick

Object removal is almost always worth doing before upscaling and final sharpening, not after: then the filled area goes through the same finishing as the whole frame and doesn't stand out. And one more: if the fill came out uneven the first time — don't redraw everything, select only the problem fragment and repeat the inpainting on that spot.

What's next

That wraps up the section on working with photos. Next come practical "for business" scenarios, and we'll start with logos.

In the Twelver chat a photo for retouching is sent right into the conversation. A few operations are free after signing up.

Try it yourself

Everything in this guide runs inside Twelver

One chat for text, images, video, music and voice — no separate services or subscriptions.

Open Twelver chat

Related pageImage Generation

Оцените свой опыт