Choose a photo and Photopip finds the faces on your device, then fills an oval over each one with solid black. A block replaces those pixels; a mosaic or a blur can be matched back to the person by software, so this page uses a block. The Rectangle Tool stays selected with a black fill, so a face the detector missed takes one drag to cover. Download a flattened copy from the card.

1. Click Choose a photo, or drop one on the page. The first time, a 232 KB face detector downloads; after that it runs on your device.
2. Each face it finds gets a black oval that also covers the forehead and hair, and the card counts them. For the example photo it reads Found and covered 3 faces. If you undo the blocks, the card says so and turns the download off.
3. Zoom in and check every face. In the example, the man far back by the purple building is too small for the detector. The Rectangle Tool is already selected with a black fill: drag a box over his face and the card updates.
4. Click Download JPG (or PNG, matching the file you opened). The file from the card is flattened: the blocks are in the pixels, with no layers and no EXIF.

The detector looks at the photo shrunk to 640 px on its long side and finds faces from about 10 px across at that size. In a photo 1,600 px wide, faces under about 25 px are often missed, and so are faces turned away and faces in reflections: check every face and drag a box over the rest. The block still shows that someone was there. One photo at a time; no batches or video. If you close the card (✕) and save a PSD instead, the boxes you dragged stay as separate layers that anyone with the file can hide; the download from the card is flattened.

Why did it miss a small face in the background?

The detector sees the photo shrunk to 640 px on its long side, so in a large photo the faces far back are only a few pixels wide by the time it looks. Faces turned away and reflections get missed too. Drag a box over anything it skipped; the Rectangle Tool is already set up for it.

Can a blurred or pixelated face still be recognized?

It can. Researchers have trained neural networks to match mosaicked and blurred faces to photos of the people (McPherson, Shokri and Shmatikov, 2016). A solid block replaces those pixels in the file you download, which is why this page uses one.

Is the photo uploaded?

No. The detector downloads once and runs in your browser, and the file is written there too. This page doesn't add the photo to Open Recent or keep an autosave copy of it.

Updated 2026-10-24

More toolsBlur a license plateRedact a screenshotCrop image

Choose a photo and Photopip finds the faces on your device, then fills an oval over each one with solid black. A block replaces those pixels; a mosaic or a blur can be matched back to the person by software, so this page uses a block. The Rectangle Tool stays selected with a black fill, so a face the detector missed takes one drag to cover. Download a flattened copy from the card.

1. Click Choose a photo, or drop one on the page. The first time, a 232 KB face detector downloads; after that it runs on your device.
2. Each face it finds gets a black oval that also covers the forehead and hair, and the card counts them. For the example photo it reads Found and covered 3 faces. If you undo the blocks, the card says so and turns the download off.
3. Zoom in and check every face. In the example, the man far back by the purple building is too small for the detector. The Rectangle Tool is already selected with a black fill: drag a box over his face and the card updates.
4. Click Download JPG (or PNG, matching the file you opened). The file from the card is flattened: the blocks are in the pixels, with no layers and no EXIF.

The detector looks at the photo shrunk to 640 px on its long side and finds faces from about 10 px across at that size. In a photo 1,600 px wide, faces under about 25 px are often missed, and so are faces turned away and faces in reflections: check every face and drag a box over the rest. The block still shows that someone was there. One photo at a time; no batches or video. If you close the card (✕) and save a PSD instead, the boxes you dragged stay as separate layers that anyone with the file can hide; the download from the card is flattened.

Why did it miss a small face in the background?

The detector sees the photo shrunk to 640 px on its long side, so in a large photo the faces far back are only a few pixels wide by the time it looks. Faces turned away and reflections get missed too. Drag a box over anything it skipped; the Rectangle Tool is already set up for it.

Can a blurred or pixelated face still be recognized?

It can. Researchers have trained neural networks to match mosaicked and blurred faces to photos of the people (McPherson, Shokri and Shmatikov, 2016). A solid block replaces those pixels in the file you download, which is why this page uses one.

Is the photo uploaded?

No. The detector downloads once and runs in your browser, and the file is written there too. This page doesn't add the photo to Open Recent or keep an autosave copy of it.

Updated 2026-10-24

More toolsBlur a license plateRedact a screenshotCrop image