How accurate is the face detection?

Very. AWS Rekognition's DetectFaces returns a tight bounding box and 27 landmarks per face with confidence usually >99% on clear front-facing portraits. Side profiles and crowd scenes are weaker — the editor lets you draw a region manually when that happens.

Can I use this for kids' photos before posting?

That's one of the most common use cases. Drop the photo, every face is highlighted, blur the ones you want hidden, leave the rest, export. No account, no record of the photo after an hour.

Does Blurify store the faces it detects?

No. We never store face descriptors, embeddings, or any biometric data. The bounding boxes live in your browser tab only. The image itself is auto-deleted from S3 within an hour.

Will the blur survive screenshot un-blurring tools?

Yes — we apply a destructive Sharp-pixel blur to the final JPG, not a reversible CSS filter. Once exported, the original pixels in the blurred region are mathematically replaced.

Blur looks too soft — can I use blocks instead?

Yes — the Pixelate style censors with hard blocks, and it has its own walkthrough: see Pixelate a face. Same detection, different statement.

What formats and sizes are supported?

JPEG, PNG, WebP and HEIC/HEIF, up to 20 MB per photo. Exports re-encode to your chosen format with metadata stripped.

Is the original photo kept anywhere?

It sits in encrypted storage just long enough to process, then auto-deletes within an hour. No account, no library, no history.

Is there a face blur app for iPhone or Android?

You don't need one — Blurify runs in the mobile browser on both iPhone and Android, with touch controls for adjusting regions. Nothing to install, no account, and the export carries no watermark.

How do I make a face unrecognizable?

Cover the whole face — hairline to chin — and don't be shy about the region size. A strong gaussian blur works; hard pixel blocks (see Pixelate a face) read as more deliberate; Solid fill is the only style that leaves nothing to reconstruct. If you can still guess the expression, go stronger.

Is blurring faces enough for GDPR anonymization?

Blurring and pixelation are widely used to reduce identifiability in published images, and Blurify's export also strips the metadata that can place a photo. But anonymization is judged by outcome, not by tool — context like clothing, location, or captions can still identify someone, and this isn't legal advice. For the strongest region treatment, use Solid fill.

Blur faces in photos online free

Open the editor How it works

Group of four friends at a cafe with every face auto-blurred — Blurify result — Rekognition finds every face — you decide which to blur, which to keep.Photo · Unsplash

Group of four friends at a cafe with every face auto-blurred — original — Rekognition finds every face — you decide which to blur, which to keep.Photo · Unsplash

How it works for faces.

Drop the photo
JPEG, PNG, WebP, HEIC/HEIF — up to 20 MB. Sent directly to encrypted storage, never to a third-party server.
Faces appear, outlined
AWS Rekognition finds every face with a tight bounding box. You see them as outlined regions on the canvas. Toggle any off.
Choose your blur, export
Gaussian, pixelate, solid fill, or color overlay. Live preview. Download with EXIF and GPS metadata stripped.

Questions, honestly answered.

How accurate is the face detection?: Very. AWS Rekognition's DetectFaces returns a tight bounding box and 27 landmarks per face with confidence usually >99% on clear front-facing portraits. Side profiles and crowd scenes are weaker — the editor lets you draw a region manually when that happens.
Can I use this for kids' photos before posting?: That's one of the most common use cases. Drop the photo, every face is highlighted, blur the ones you want hidden, leave the rest, export. No account, no record of the photo after an hour.
Does Blurify store the faces it detects?: No. We never store face descriptors, embeddings, or any biometric data. The bounding boxes live in your browser tab only. The image itself is auto-deleted from S3 within an hour.
Will the blur survive screenshot un-blurring tools?: Yes — we apply a destructive Sharp-pixel blur to the final JPG, not a reversible CSS filter. Once exported, the original pixels in the blurred region are mathematically replaced.
Blur looks too soft — can I use blocks instead?: Yes — the Pixelate style censors with hard blocks, and it has its own walkthrough: see Pixelate a face. Same detection, different statement.
What formats and sizes are supported?: JPEG, PNG, WebP and HEIC/HEIF, up to 20 MB per photo. Exports re-encode to your chosen format with metadata stripped.
Is the original photo kept anywhere?: It sits in encrypted storage just long enough to process, then auto-deletes within an hour. No account, no library, no history.
Is there a face blur app for iPhone or Android?: You don't need one — Blurify runs in the mobile browser on both iPhone and Android, with touch controls for adjusting regions. Nothing to install, no account, and the export carries no watermark.
How do I make a face unrecognizable?: Cover the whole face — hairline to chin — and don't be shy about the region size. A strong gaussian blur works; hard pixel blocks (see Pixelate a face) read as more deliberate; Solid fill is the only style that leaves nothing to reconstruct. If you can still guess the expression, go stronger.
Is blurring faces enough for GDPR anonymization?: Blurring and pixelation are widely used to reduce identifiability in published images, and Blurify's export also strips the metadata that can place a photo. But anonymization is judged by outcome, not by tool — context like clothing, location, or captions can still identify someone, and this isn't legal advice. For the strongest region treatment, use Solid fill.