Image Blur / Sharpen
Gaussian blur or unsharp-mask sharpen.
Overview
The Image Blur / Sharpen tool applies a Gaussian blur or unsharp-mask sharpening pass to an uploaded photo, with a slider that controls the kernel radius. Upload a picture, pick a mode and a strength, and the tool returns a re-encoded copy ready for download.
Blurring is the right call for background plates, screenshot redaction, and stylised social posts; sharpening lifts the perceived crispness of slightly soft scans and downsized exports. Both effects are common requests in a workflow that asks for a Gaussian blur on a photo or an unsharp mask sharpen pass online.
How it works
Gaussian blur convolves the image with a 2-D Gaussian kernel G(x,y) = (1 / (2 pi sigma^2)) * exp(-(x^2 + y^2) / (2 sigma^2)). In practice the operator is separated into a horizontal pass followed by a vertical pass so the cost grows linearly with kernel size rather than quadratically. The slider controls sigma; the kernel radius is typically clamped to roughly 3 sigma.
Unsharp masking works the opposite way: the image is blurred at a chosen radius to produce a low-frequency reference, that reference is subtracted from the original to leave only the high-frequency detail, and the detail is added back at a configurable amplitude. The net result emphasises edges without inventing new ones, which is why the technique survives from darkroom printing to modern raw processors.
Examples
Before: portrait.jpg (sharp 4000x6000 photo)
After: portrait-blur.jpg with sigma=8 → soft bokeh-style background
Before: scan.jpg (slightly soft after downsizing to 1200 px)
After: scan-sharpen.jpg with radius=1, amount=80 → crisp edges
Before: screenshot.png with sensitive text
After: screenshot-blur.png with sigma=12 → text unreadable
FAQ
What sigma do I need for redacting text?
Match sigma to roughly half the character height in pixels — for 24-pixel-tall text a sigma of 10-14 makes the characters unreadable. Verify by reading the result back.
Is unsharp masking the same as sharpening?
Yes — almost every "sharpen" slider in editors is some variant of unsharp masking. Only the kernel and clamping differ between implementations.
Why does blurring darken the image edges?
If the implementation clamps at the boundary instead of extending pixels, edges receive fewer samples and appear darker. The tool extends border pixels to avoid this artefact.
Can I blur only part of an image?
Not in this tool — it processes the entire frame. Crop first or composite the blurred output with the original in a separate editor.
Does sharpening introduce noise?
It amplifies whatever high-frequency content is already there, including sensor noise. Keep the amount under 100 for noisy photographs.