Hide faces before you post — protect bystanders, children, or anyone who didn't agree to be shared. Find the faces automatically, check the boxes, and download. The detection runs on your device too, so the photo never leaves it.
Your files never leave this deviceJPEG, PNG, WebP · up to 30 images at once · 50 MB each
Four steps, a few seconds, zero uploads.
Drag your images onto the editor, or click Browse. JPEG, PNG, and WebP are all accepted, up to 30 at a time.
Click Find faces to detect them automatically across every photo. Then check its work: drag a box over anything it missed, and move, resize or delete any box you don't want.
Raise the strength until no facial features remain — use the strongest setting for small or distant faces.
Save the result. The blur is rendered into the pixels, so the hidden face can't be recovered.
Blurring a face is usually about privacy — a stranger in the background, a child, a colleague, or a witness who shouldn't be identifiable. It makes little sense to hand that same photo to a remote server just to hide it. Latch does the whole job in your browser: the image is decoded, edited, and re-encoded on your own device, and no copy is ever sent anywhere.
The result matters too. Some tools draw a blur as a separate layer that can be peeled back from the file. Here the blur is composited straight into the exported pixels at full resolution, so once you share the image there's no sharp original hiding underneath. For anything sensitive, turn the strength up until the face is completely unreadable.
Automatic face detection normally means an upload: you hand the photo to a server, it finds the faces, and it sends coordinates back. That's a strange trade for a privacy tool — you've disclosed the very image you were trying to protect, to a company whose retention policy you have to take on trust.
This page does it differently. Clicking Find faces downloads a small open-source detection model (YuNet, from the OpenCV Zoo) into your browser once, and every photo after that is scanned locally. Nothing about the image — not the pixels, not the coordinates, not a thumbnail — is transmitted. It works across a whole batch too, so 30 photos are scanned one after another without any of them leaving your device.
The honest caveat: on-device detection is a first pass, not a guarantee. It reliably finds faces that are reasonably large and facing the camera, and it will miss some that are small, turned away, or in shadow. Treat the boxes it draws as a starting point and check every photo before you share it — that's why each box stays editable, and why nothing is ever exported until you press the button yourself.
No. The photo is decoded and edited entirely in your browser using the Canvas API. Nothing is sent to a server — that includes the face finder, which downloads once and then runs on your own device. Editing and export keep working even if you go offline after loading the page.
No — and it's worth being clear about that. The finder is a helper, not a guarantee. It does a first pass and is good at faces that are reasonably large and facing the camera, but it can miss faces that are small, turned away, in deep shadow, or partly covered. Always check the result and draw a box over anything it missed before you share the image.
No. The detection model is downloaded to your browser the first time you use it, then runs entirely on your device. Most tools that offer automatic face blurring upload your photo to a server to do the detection — this one never does.
No. The blur is rendered into the actual pixels of the exported image at full resolution — it is not a layer sitting on top. Once you download and share the result there is no original detail underneath to reverse. For sensitive cases, use a high strength so no facial features remain.
Turn the strength up until no facial features are recognisable — for a small or low-resolution face, use the strongest setting. Prefer mosaic blocks or a solid bar instead of a soft blur? The pixelate and redact tools do the same job with a different look.
Yes. Draw a separate box over each face — there is no limit. You can move, resize, or delete any box before you export.
Yes. The editor supports touch, so you can draw and adjust boxes with your finger on a phone or tablet just as you would with a mouse.
Related privacy tools: Blur part of an image · Pixelate an image · Blur a license plate · Redact an image · Remove photo metadata
How much blur is enough? What a blur does and doesn't destroy — strength is what decides it.