Unsharp Masks are Cool
No, not the masks that you wear to help fight the spread of coronavirus, I'm talking about the masks that help your images to look sharp. They may not sound sharp, but they are definitely cool.
First, it's worth considering why images need sharpening at all? It's tempting to assume that if you do everything right when you shoot an image, it should be sharp right out of the camera. Focus carefully, and your job is done, right? But such is not entirely the case. Forcing your view of the world to conform to a regular grid of pixels means everything has to fit within those tiny boxes. Things ends up at least a tad bit approximated, with each pixel being an average of what lies inside. As camera resolution increases, each pixel becomes correspondingly smaller, but the problem remains the same if you look closely enough. A small amount of sharpening restores the appearance of detail lost during capture and processing. Within reason, a bit of sharpening can help save a soft image caused by other reasons, too.
But unsharp masking can seem a bit of a counterintuitive enigma. For something intended to improve detail rendering, it sure has a funny name. One would think it should be called a "sharp mask," at least. To understand why a cool sharpening technique has such a dull name, we need to look at some history.
Digital sharpening has the advantage of operating in software, and software can do all sorts of things. All someone has to do is figure out the math of how to manipulate pixel color values. But film photographers had a similar problem, with the outside world mapped to random particles of film grain rather than pixels. Fundamentally, any image capture method causes a degree of, and while film and digital are quite different, they both can benefit from a bit of sharpening.
With film, the means to address the problem were limited. Back in the old days, the process went something like this. Imagine you created a negative copy of your original image. If you then sandwich this in with the original, the two would cancel each other out. The dark parts of the copy would counterbalance the lighter areas of the original, and vice versa. But now imagine that the negative is rendered slightly out of focus. The light and dark would blur together somewhat near any edges. When stacked together with the original, this blurred negative will now accentuate detail by diminishing the canceling effect near edges. If one side of a border in the image was dark gray and the opposite one light gray, the in-focus negative would reverse the tones accurately, and everything simply would be dimmed. With a blurred negative, the two shades of gray would mix along the edges, and the boundaries would stand out more clearly than the rest. With appropriate exposures and blending, this is the gist of traditional darkroom unsharp mask.
Today we work In the digital darkroom, and software doesn't require optical and mechanical means to produce results. All it needs is some math and a fast CPU processor.
When you open the Unsharp Mask dialog in almost any software program, you will see pretty much the same thing, but I'm going to use the terminology here from Photoshop. If you're image editing software differs, you probably won't have too much trouble following along. The Amount slider controls how strong the effect will be. The Radius slider determines the size of the net you will be casting to find edges. Anything further away than this value from each given point will not be an influence in considering that point an edge. Set this too small, and only the finest edges will fit within the Radius and be detected. Set high enough, and random dust spots, and other irrelevant details could be considered part of some "edge." Think of this setting as being analogous to the blur in the film darkroom version. Threshold allows you to weed out some false matches for edges and is especially useful when working on images with noise or dust. The last thing you want is to sharpen the appearance of dust spots. Raise this slider until small speckles pass safely without being sharpened. But on the downside, higher Threshold values could also result in finer details going undetected and thus unsharpened.
With three independent control sliders, learning to use Unsharp Mask can take some practice. First, make sure you are not sharpening your original directly. Copy the layer, or convert it to a Smart Object. One thing about sharpening is that it can be tempting to overdo it. You want to leave yourself the option to dial things back or otherwise tweak things the next day. When viewed afresh, you may have second thoughts.
Not every image will require the same Unsharp Mask settings. Depending on your camera and your preferences, you may find values that work well as a starting point, but to get the best results, you want to eyeball it. Zoom in on your image enough to tell what you are doing, but not all the way. You don't want results that can only be seen under a microscope. It used to be standard advice to zoom to 100%, but with new, higher resolution cameras, I find that this is overkill. Keep the Preview checkbox enabled, and watch what you are doing. If you overdo it, the exaggeration of contrast that makes Unsharp Mask work will become obvious, and you will begin to see halos around edges. There are no right answers, but use restraint. Nothing looks worse than an over-sharpened image.
Creative folks have developed several other sharpening filters, but the unsharp mask remains the standard by which all others are judged. They all rely, at least in part, on masking.
So, mask up, everybody.