Unsharp mask, abbreviated USM, is the classic image sharpening technique found in virtually every editing program. Despite its confusing name, it makes images appear sharper. It works by increasing contrast along edges, making the light side of an edge a little lighter and the dark side a little darker, which the eye reads as crisper detail even though no genuinely new detail is created.
The name comes from a darkroom process. Photographers created a blurred, or unsharp, negative copy of an image and sandwiched it with the original during printing, and the combination of the sharp original with its blurred mask enhanced edge contrast. Digital unsharp mask reproduces this mathematically by blurring a copy of the image and comparing it to the original to find the edges, then boosting the difference between them.
Three controls govern the effect. Amount sets the strength of the contrast boost, how much lighter and darker the edges become, often in the range of 80 to 150 percent. Radius sets how wide a band around each edge is affected, with small radii near one pixel sharpening fine detail and large radii adding broad contrast. Threshold sets how different two areas must be before the filter treats the boundary as an edge.
Threshold is the control that protects smooth areas. Raising it tells the filter to ignore low-contrast differences, so gentle gradients such as skies and skin are left alone while strong edges are still sharpened, which prevents noise in flat areas from being amplified into something ugly. Setting it to zero sharpens everything, including grain.
The main risk is overdoing it, which produces halos, visible light and dark outlines around high-contrast edges, along with a crunchy, artificial look and exaggerated texture. Subtlety wins, and many editors apply sharpening through a mask so it strengthens detailed areas like eyes and edges while leaving smooth ones untouched. Viewing at 100 percent while adjusting is essential to judge the effect honestly.
Sharpening is best thought of in stages. A small amount of capture sharpening offsets the softening introduced by the sensor’s demosaicing and any anti-aliasing filter, an optional creative pass adds local punch, and a final pass of output sharpening is tuned to the medium, since a large print and a web image need very different treatment. Sharpening is always the last step before export, applied once the image is at its final size, because resizing a sharpened image reintroduces softness or artifacts.
A safe starting recipe for output sharpening is a small radius near one pixel, a moderate amount, and a low threshold raised just enough to leave skies clean, then judged at the final viewing size. When in doubt, less is more, because under-sharpening merely looks soft while over-sharpening looks broken and cannot be undone after export.