Blur Image Online

Blurring helps protect privacy, soften backgrounds, add depth-of-field style, and guide attention—without installing desktop software.

Benefits

• Privacy: Obscure faces, plates, or sensitive details before sharing.
• Background softening: Push backgrounds back so subjects read clearer.
• Creative look: Soft focus, motion, or bokeh-style effects.
• Focus: Reduce visual noise in busy areas.
• Polish: Quick finish for social and web assets.

1. 1Upload your image: JPG, PNG, WEBP, GIF, and BMP. You can add several images for batch processing.
2. 2Choose blur type and strength: Pick Gaussian, box, motion, radial, or lens blur. Move the intensity slider; preview updates live.
3. 3Preview the result: Check the blurred preview. Adjust until the look matches your goal—privacy, soft focus, or creative effect.
4. 4Export: Download as PNG (transparency), JPEG, or WEBP.

A blur is defined by its shape — the pattern of neighbouring pixels each output pixel is averaged from. A Gaussian blur uses a shape that fades smoothly from the middle outward, so a bright point of light becomes a soft, formless smudge. A camera lens does something quite different: it spreads each point into a *disc*, flat and evenly bright right up to a hard round edge. That is why out-of-focus highlights in a photograph appear as distinct circles, and it is the whole visual signature of a fast lens. Measured on the tool this replaces, its "Lens Blur" differed from its own plain Gaussian blur by a maximum of 2 units out of 255 — the two were the same operation with a different name in the menu. The bokeh here is a real disc: on an isolated highlight it holds 77% of its peak brightness out to 80% of its radius and falls to nothing past the edge, where a Gaussian at the same setting has already dropped to 25%. Turn the amount up on a photo with points of light behind the subject and the difference is not subtle.

Both turn on a centre point, and they are frequently confused because photo editors have historically bundled them under one name. A spin blur smears each pixel along the *arc* it would travel if the picture rotated — the effect of a spinning wheel, or a camera twisted on its axis. A zoom blur smears each pixel along the *radius*, away from the centre — the effect of pulling a zoom lens during the exposure. They look nothing alike, and they preserve opposite things: spin keeps a circle drawn around the centre intact and destroys a line running outward from it; zoom does exactly the reverse. The tool this replaces offered one control labelled "Radial Blur — Circular blur from centre" whose code performed the zoom. Tested on an image with both a ring and a radial spoke, the ring was smeared from 255 down to 57 while the spoke came through untouched — the label described one blur and the tool delivered the other. Both are here now, separately, each with a movable centre.

Blurring an entire photograph is rarely the goal. Usually you want the subject sharp and the background soft, or a narrow band in focus and everything above and below it melting away, or a face hidden while the rest of the scene stays readable. All three are the same mechanism: a mask that says how much blur each part of the picture gets. A round shape keeps the middle sharp and softens what surrounds it, which is background blur; a straight band does the same along a line, which is the tilt-shift look; and marked rectangles blur what you cover rather than what you leave, which is redaction. Every one of them can be swapped, so the round shape can blur the middle and leave the surroundings sharp instead. Turn on Show where it applies and the picture is replaced by the mask itself: white gets blurred, black is left alone. The tool this replaces could only blur inside a rectangle — there was no way to blur *around* anything — while shipping a preset called "Portrait Background" that was a plain Gaussian over the whole frame, subject included. There is also a fourth option that finds the subject for you and blurs everything around it, using the same detection the background remover runs — useful when the subject is off-centre or an awkward shape that no circle fits. It takes a moment the first time on each image, and only loads anything at all if you choose it.

A mosaic replaces each block of the picture with a single averaged colour. Visually it is cruder than a blur, and that is exactly why it is the better choice for hiding something: a blur looks like a softening effect that might be reversible, while blocks read unmistakably as information that has been deliberately removed. Use it for licence plates, screen contents, addresses and documents. Two practical notes. Larger blocks hide more — a block smaller than the text you are covering can leave the shape of individual letters legible. And whatever you use, the pixels are genuinely discarded here rather than covered over: the exported file contains the blocks, not the original with something drawn on top. If you specifically want faces found for you, the face blur tool does the detection and this one does not.

Transparent images carry a colour for every pixel, including the pixels you cannot see, and in most files that hidden colour is black. Blur those pixels together with the visible ones and the invisible black bleeds inward, so a soft-edged logo or cut-out picks up a grey rim that was never in the original. The fix is to weight every pixel's colour by how visible it is before averaging, so a fully transparent neighbour contributes nothing rather than contributing black. It is a well-understood step and it is not optional; measured on the tool this replaces, a pure white shape over transparency came back at 42 out of 255 along its visible edge — an 84% darkening. Here it stays white, and the test that proves it also runs the arithmetic the old way to confirm the halo really does appear without the fix.

Blur filters average nearby pixels so edges soften. Gaussian blur uses a smooth weight curve; the tool also offers box, motion, radial, and lens-style blur for different looks.

Gaussian blur

A Gaussian kernel weights neighbors by distance, producing smooth, natural softness. Higher strength uses a wider neighborhood for stronger blur.

Processing steps

• Load Image data is read straight from the file.
• Kernel Blur radius and type define how pixels are mixed.
• Convolution Each channel is filtered (with edge handling).
• Preview The preview shows the result before you export.
• Encode Save as PNG, JPEG, or WEBP.

Match strength to your goal—subtle softening vs strong privacy or abstract effects.

Typical ranges

• Subtle (about 1–3) Light softening; detail mostly intact.
• Moderate (about 4–8) Clear background softening and depth-style effects.
• Strong (about 9–15) Heavier blur for privacy or bold creative use.
• Very strong (16+) Strong obscuring or abstract look—depends on image size.

A blur touches every pixel and averages it with everything nearby, so the work grows with the size of the picture and with the amount of blur together — a heavy blur on a large photo is a very large number of additions. Done in ordinary code, a 12 megapixel photo at a moderate setting takes about five and a half seconds, which is why the tool this replaces felt broken: it re-ran that full-resolution calculation every time a slider moved, and re-encoded the file afterwards, while describing itself as a real-time preview. This tool hands the work to your graphics hardware where it is available, which brings the same job to about a seventh of a second, and previews at screen size rather than full size so dragging a slider stays immediate. Pictures too large for a single pass are split into overlapping tiles and rejoined with enough overlap that the seams cannot appear — the old tool simply refused anything above 16 megapixels, which ruled out most current cameras.

Blur shows up in privacy workflows, photography-style edits, social posts, and web design.

• Background remover
• Image compressor
• Image sharpener