What it does
Applies an unsharp-mask filter to your image, sharpening edges and adding clarity. The strength slider controls how aggressive the sharpening is, lower values are subtle, higher values are dramatic. Most photos look best with a sigma value between 0.7 and 1.5.
Server-side processing uses libvips through Sharp, the same image library that powers most professional image pipelines.
What “sharpening” actually does
Unsharp mask is a counterintuitive name for a long-standing technique: subtract a blurred copy of the image from the original. The difference is the high-frequency detail (edges, fine texture). Adding that difference back to the original amplifies the edges, making the image look sharper.
This works best on slightly soft images, a photo from a phone camera, a JPEG that’s been compressed, or anything captured at the limits of your lens. Sharpening can’t recover detail that was never captured. If your photo is severely blurry or out of focus, sharpening just amplifies noise.
When to use what amount
- 0.3-0.7: very subtle, good for already-sharp photos that just need a tiny boost
- 0.8-1.2: typical default, works well for most casual photos
- 1.5-2.5: aggressive, useful for soft web exports or recovering slightly out-of-focus images
- 3.0+: too much for most cases, produces visible halos around edges and amplifies noise
The slider lets you find the sweet spot for your specific image. The same value won’t work for every photo, landscapes need different treatment than portraits.
Frequently asked questions
Why is the sharpened image larger than the original? Sharpening preserves more high-frequency detail, which JPEG/WebP encoders then have a harder time compressing. The output file is typically 5-15% larger than the input. If size matters, run the result through the image compressor afterward.
Can I sharpen a JPEG without quality loss? JPEGs are lossy by nature, any re-save introduces some compression loss. The sharpening filter itself is lossless, but the re-encoded JPEG won’t be bit-identical to a hypothetical lossless sharpened image. For pixel-perfect work, use PNG output.
Does this work on transparent PNGs? Yes. The alpha channel is preserved.
Why does heavy sharpening look “fake”? Unsharp mask amplifies edges symmetrically, same sharpening on both sides of an edge. Real lens optics don’t produce perfectly symmetric sharpening. Past about 2.5 sigma, the result starts looking computer-generated. Subtle is usually better.