How to use Face & text blur
- 1
Add photos
Drop up to three photos or screenshots; faces are detected the moment each one loads.
- 2
Find text
Click Find text to cover licence plates, names, numbers and other words.
- 3
Adjust the areas
Untick anything to keep, and add rectangles, ellipses or brush strokes over the rest.
- 4
Download
Choose Blur, Pixelate or Solid box and save the image, or get every image as a ZIP.
Limits and Pro
Up to 3 images per batch, each up to 25 MB, with every effect, manual shapes and the ZIP download.
Batches of up to 500 images for automatic face blurring, plus files larger than 25 MB. See Pro
About Face & text blur
Face & text blur produces a copy of your image in which chosen areas are painted over with a strong blur, a mosaic of pixel blocks or a solid colour box. Everything outside those areas keeps its original pixels, and the export keeps the original resolution and format — JPG stays JPG and PNG keeps its transparency — while camera metadata such as GPS coordinates is dropped because the file is re-encoded.
Faces are found by a BlazeFace detector that scans the whole picture and overlapping tiles, so small faces in crowds and group photos are caught as well. Text detection uses OCR to locate words and merges them into line-shaped boxes. Each detection appears in a list you can switch on or off, every shape can be moved or resized with handles, and a freehand brush covers awkward objects such as tattoos, house numbers or papers lying on a desk.
Typical uses are anonymising bystanders before posting street photography, hiding children's faces, redacting names and email addresses from screenshots, and covering licence plates. Detection is not perfect — profile views, deep shadow, tiny print and stylised lettering can be missed — so look over the preview before sharing. For passwords, card numbers and ID documents choose Solid box, since blur and coarse pixelation can occasionally be partly reversed.
Privacy
Face detection, OCR and the blurring itself are computed on your device, so your file never leaves your device — which matters most when the whole point is protecting the people in the picture. The detector model and OCR language data come from public CDNs when first needed; the images you edit are never uploaded or stored.
Features
Automatic face detection
A multi-scale scan catches small faces in crowds and group shots.
Text detection
OCR finds plates, badges, emails and street names and boxes each line.
Shapes and brush
Rectangles, ellipses and freehand strokes with move and resize handles.
Three redaction styles
Blur, pixelate or a solid box, with adjustable strength.