Blur Image Online: Blur Photos, Backgrounds and Faces Free

What is Blur Image Online?

Blur Image Online is a free tool from Entertainment Couch that blurs photos in your web browser, with no sign-up, upload or watermark. You can blur the whole image, the background behind a person, every face in a group photo, or any part you select, using Gaussian, pixelate, motion, radial, lens and tilt-shift effects, and you can cover private text with a solid redaction block. The blur is written into the pixels of the file you download.

Blur Image Online at a glance

How do you blur an image online?

Add your photo, choose what to blur (whole image, background, faces or part of the image), set the blur intensity with the slider, check the result with Compare, and press Download. The edited file keeps the original size, and nothing is uploaded.

1. 1

   Upload your image

   Press Choose images, drag a photo onto the tool, or paste a screenshot with Ctrl+V (Cmd+V on a Mac).

2. 2

   Choose what to blur

   Pick Whole image, Background, Faces, Part of image, Hide text, Tilt-shift or Focus point, or draw straight on the photo.

3. 3

   Adjust the intensity

   Move the Blur intensity slider, pick a blur type, and soften the edge with Feather. The preview updates as you go.

4. 4

   Preview and download

   Press Compare to drag a before-and-after divider, then press Download to save the blurred image.

How do you blur the background of a photo?

Add the photo and press Background. An on-device AI model finds the people in the picture, keeps them sharp and blurs everything behind them. Use the Eraser to keep more of the scene sharp, or the Brush to blur more. For a camera-like look, choose the Portrait preset, which uses lens blur that gets stronger with distance from the subject.

1. Find the subject: the first time, the browser downloads a small AI model (about 5 MB) that separates people from the background, including hair, without sending your photo anywhere.
2. Blur the background: the background is blurred with Gaussian blur at intensity 45. Change it with the Blur intensity slider, or switch to Lens for soft bokeh highlights.
3. Keep the edges clean: Edge-aware blurring stops the subject's colours leaking into the blurred background, which is what causes a bright halo around a person in simpler tools. Snap mask edges makes the outline follow the edges in the photo.
4. Fine-tune by hand: paint with the Eraser over anything that should stay sharp, such as a bag, a pet or a product the person is holding.
5. Simulate depth: Gradual blur with distance blurs the background lightly next to the subject and more strongly further away, which looks closer to a real wide-aperture lens.

The AI recognises people. For a pet, a car, food or a product photo, choose Whole image or Focus point and paint the subject with the Eraser, with Snap mask edges turned on.

How do you blur faces in a photo?

Press Faces. The face detector scans the whole photo, including small faces at the back of a group, and blurs every face it is confident about. Each face appears in the Faces panel with a tick box, so you can blur all of them, only some of them, or add a face that was missed by drawing an ellipse over it.

• Choose the effect for faces: Blur, Pixelate, or Solid block. A solid block is the only choice that hides a face for certain.
• Strength and area: the Strength slider changes how strongly every face is blurred, and Area around each face widens the covered oval to include hair and ears.
• Possible faces: detections the model is unsure about are shown with a dashed amber outline and are not blurred until you tick them. Click one on the image to include it.
• One face at a time: click a face on the image to select it and give it its own blur type and strength.

Face detection works best on faces that look roughly towards the camera and are at least about 20 pixels wide. Faces in profile, heavily shadowed, very small or partly covered can be missed, so always check the result before you share it.

How do you blur part of an image?

Choose Box, Ellipse or Freehand and drag over the area you want to blur, or use the Brush to paint blur onto exactly the pixels you want. Each area you draw gets its own blur type, intensity and feather, so one photo can have several blurred areas with different strengths.

• Box and Ellipse: drag to draw, then drag the round handles to resize or the shape itself to move it. Hold Shift for a square or circle.
• Freehand: trace any outline. The shape closes when you let go.
• Brush: set the size, hardness and opacity, then paint. Switch to Remove blur (the Eraser) to take blur away again.
• Invert: tick Invert to blur everything except the area you drew, for example to keep one product sharp in a busy shot.
• Feather: soften the transition between blurred and sharp so the edit is not noticeable.

How do you hide text, numbers and other private information in an image?

Press Hide text / info and drag a box over the text. Redact replaces every pixel inside the box with a solid colour in the downloaded file, so no part of the original remains. Do not rely on light blur or large pixels to hide names, account numbers, addresses or ID numbers, because researchers have shown that blurred and pixelated text can often be read back.

Blur and pixelation keep a version of the original information: pixelation stores the average colour of each block, and blur spreads the original pixels around. When the possible content is limited, such as digits on a card or a known font, it can be matched by trying every possibility, a weakness demonstrated in research on mosaicked documents [5] and on blurred and pixelated faces and text using machine learning [4].

• Use Redact for text and numbers: it has no feathering and no partial edges. Any pixel even slightly inside the shape is filled.
• Redaction is applied first: other blurs never read the hidden pixels, so they cannot smear traces of them into neighbouring areas.
• Screenshots: paste a screenshot with Ctrl+V, redact it, then press Copy to paste the clean version straight into a chat or email.
• Metadata: downloads carry no EXIF data, so the camera model, date and GPS location of the original photo are not passed on.

Which blur type should you use?

Use Gaussian blur for a smooth, natural softening, Pixelate for the familiar mosaic look, Motion blur to suggest speed, Radial for zoom or spin effects, Lens blur for bokeh-style backgrounds, Tilt-shift for a miniature look, and Redact whenever information must be hidden for certain.

Blur presets

Light, Medium and Strong set a Gaussian blur of intensity 12, 30 and 65. Portrait blurs the background with lens blur that increases with distance. Background applies a strong Gaussian blur behind people. Privacy pixelates the selected area, or every detected face if nothing is selected. Custom lights up whenever your own settings differ from a preset.

Does blurring reduce image quality or resolution?

No. The downloaded image has the same width, height and aspect ratio as your original, unless you choose a smaller size, and only the areas you blurred change. Choose PNG for no further compression, or JPG or WebP at quality 92 for a smaller file that looks the same.

• Resolution: the preview may be shown at a lower resolution on slow devices, but the download is always rendered again from the full-size original with the same settings.
• Very large photos: each browser has a maximum image size it can process (often 16,384 pixels on the long side on computers and 4,096 to 16,384 on phones). Larger photos are edited at the largest size your browser supports, and the tool tells you when that happens.
• Orientation: photos taken with the phone turned sideways open the right way up, because the camera's orientation tag is applied before editing.
• Transparency: transparent PNG and WebP images stay transparent, and blur near transparent edges does not create dark fringes. Saving a transparent image as JPG fills the transparent areas with white.
• Colour: images are processed in the sRGB colour space used by the web. Photos saved in a wide colour space such as Display P3 or Adobe RGB are converted to sRGB, which can make very saturated colours slightly less vivid.
• Animated GIFs: only the first frame is used.

Is it safe to blur photos online with this tool?

Yes. Your images never leave your device: the blur engine, the face detector and the background model run inside your browser, and the finished file is created there too. Nothing is uploaded, stored or shared.

The page downloads the blur engine with the page, and downloads the AI models from a public code library the first time you use Background or Faces. After that, your photo is processed by your own device's graphics chip. You can check this yourself: open the page, turn off your internet connection, and blur a photo with any tool except Background and Faces. It still works.

Closing or reloading the page clears your images from memory. Downloaded files are re-encoded from scratch, so EXIF metadata such as GPS location, camera serial number and the original date is not included.

What are the limitations of blurring images in a browser?

Automatic background blur recognises people, not pets or objects; face detection can miss small, turned or covered faces; very large photos are limited by your browser's graphics memory; and blur or pixelation is not a secure way to hide text. Each limitation has a manual alternative in the tool.

• Subject detection: trained on people. Other subjects can be kept sharp by painting them with the Eraser and Snap mask edges.
• Faces: very small faces, side profiles, strong shadows, masks and sunglasses reduce detection. Possible faces are shown for you to confirm, and you can add faces by hand.
• Lens blur depth: the depth effect is simulated from the distance to the person, not measured. Objects in front of the subject are treated as background.
• Graphics requirements: the tool needs WebGL 2, which current versions of Chrome, Edge, Firefox and Safari support. If it is turned off in your browser settings, the editor cannot start and tells you so.
• Memory: photos above roughly 50 megapixels may be too large to download at full size on some phones. Choose 75% or 50% if that happens.
• Not a forensic redaction tool: Redact removes the pixels from the image file, but it cannot remove information visible elsewhere, such as a reflection, or text in a file name.

Key terms for blurring images

Short definitions of the words used on this page.

Gaussian blur

A blur that averages each pixel with its neighbours, weighting nearer pixels more, following a bell-shaped (Gaussian) curve.

Pixelation (mosaic)

Replacing the image with large square blocks of a single colour, each the average of the pixels it covers.

Redaction

Removing information completely, here by replacing the pixels with an opaque solid colour.

Mask

A greyscale map of where an effect applies: white is fully blurred, black is untouched, grey is partly blurred.

Feather

A gradual transition at the edge of a mask, so blurred and sharp areas blend without a visible line.

Bokeh

The look of out-of-focus areas in a photo, especially the round discs formed by bright points of light.

Tilt-shift

A technique that keeps a narrow band in focus so a real scene looks like a small model.

EXIF

Information stored inside a photo file by the camera, which can include the date, camera model and GPS location.

About Blur Image Online

Who makes the tool, how it works, and the research it is based on.

Who made it

Blur Image Online is designed and maintained by Entertainment Couch, which also publishes free browser games and a set of free online tools. It was last updated on 6 November 2026.

How it works

• Every effect is computed on your device's graphics processor with WebGL 2 [6]. Large Gaussian blurs are calculated on a reduced-resolution pyramid and smoothly scaled back, which gives the same result as a full-size blur in a fraction of the time.
• Selections are stored as shapes and brush strokes, not pixels, so every area can be changed, hidden or removed until you download. The download is rendered again at full size from the original image.
• Mask edges can be snapped to the photo with a guided filter [1]. Edge-aware background blur averages only background pixels, so the subject does not leave a halo.
• Background blur uses the Robust Video Matting model [2], and face detection uses the BlazeFace family of detectors [3], run with TensorFlow.js [7] in the TF.js conversions published as human-models [8]. The photo is scanned in overlapping tiles so small faces are found.

Sources

1. He, K., Sun, J. and Tang, X. (2013). Guided Image Filtering. IEEE Transactions on Pattern Analysis and Machine Intelligence 35(6). doi.org/10.1109/TPAMI.2012.213
2. Lin, S., Yang, L., Saleemi, I. and Sengupta, S. (2021). Robust High-Resolution Video Matting with Temporal Guidance. arxiv.org/abs/2108.11515
3. Bazarevsky, V. et al. (2019). BlazeFace: Sub-millisecond Neural Face Detection on Mobile GPUs. arxiv.org/abs/1907.05047
4. McPherson, R., Shokri, R. and Shmatikov, V. (2016). Defeating Image Obfuscation with Deep Learning. arxiv.org/abs/1609.00408
5. Hill, S., Zhou, Z., Saul, L. and Shacham, H. (2016). On the (In)effectiveness of Mosaicing and Blurring as Tools for Document Redaction. Proceedings on Privacy Enhancing Technologies 2016(4). petsymposium.org/popets/2016/popets-2016-0047.php
6. Khronos Group. WebGL 2.0 Specification. registry.khronos.org/webgl/specs/latest/2.0/
7. Smilkov, D. et al. (2019). TensorFlow.js: Machine Learning for the Web and Beyond. arxiv.org/abs/1901.05350
8. Mandic, V. human-models: pre-trained models for the Human library (MIT licence). github.com/vladmandic/human-models

More free image tools on Entertainment Couch

All free in your browser, with nothing to install.

Frequently asked questions

Search, or pick a topic. Blur strength, backgrounds, faces, privacy, quality and more.