Photo editing guides

Compare sharpening methods on a printed label

·

A crisp label includes flat areas and a boundary where the tone changes quickly. Follow that boundary in the example to see how two filters affect the same source pixels.

Image Sharpener

Check two pixels at the label edge

Load the example. The gray values on row 40 at columns 71 and 72 are 80 and 160, using zero-based coordinates. Select Unsharp Mask, radius 2, amount 100% and threshold 0. The two values become 56 and 184. The filter adds back the difference between the original and a blurred copy.

scikit-image Unsharp Mask reference

Compare the immediate-neighbor filter

Select Laplacian with amount 100% and threshold 0. Its four-neighbor measurement changes those same values to 0 and 240. This gives a stronger change at the boundary, with a dark side already at the lowest possible 8-bit value.

OpenCV image filtering reference and Laplacian kernel

Reduce visible outlines

Try amount 50% to halve the added edge difference. Increase the threshold if small variations in blank paper begin to stand out. Compare the edge and the nearby gray area, then save the chosen result.

Common questions

Which filter uses the radius setting?

Unsharp Mask uses the radius of its Gaussian blur. The Laplacian setting always uses the four immediate neighbors.

Why use a higher threshold on a noisy scan?

A higher threshold leaves more small differences alone. Compare the grain in flat areas and the lettering after each adjustment.

Sources

Put it into practice.

Related guides