All digital sharpening is carried out on pixels. Throughout this book, we'll examine images at the pixel level to see what happens to the pixels through various stages of image reproduction. But with the exception of images destined for on-screen display on the Web or in an information kiosk, the final product is not pixels, but marks on paper or some other substrate.
But the relationship between the image pixels and the marks on paper is rarely obvious. In conventional halftone printingthe kind we use on printing presses, it typically takes between two and four pixels to form a single halftone dot in grayscale printing. When we print in color, the relationship is even less obvious since the halftone dots are spread across four different color plates.
Inkjet printers use a different type of screening, but again, the relationship between printer dots is indirect. Even in the relatively simple scenario of printing to a true continuous-tone printer, where the pixels are reproduced one-to-one, the pixels on the print are typically much smaller than the same pixels on screen.
How Big are Your Pixels?
We can't control how the pixels get turned into printer dots. We can control the degree of contrast boost, and the size of the pixels on output. In so doing, we control, in turn, the width of the sharpening "halo." The goal is to avoid haloes that our eyes pick out as discrete features, while making them strong enough, and (just) wide enough to produce a sharpening effect.
Final Sharpening must be Done at Final Output Size and Resolution
Since we can only control the final size of the pixels, it follows that we must do our final sharpening at final output size and resolution if we want to get the sharpening halo right. If the image is resized after sharpening, the haloes either get bigger, and become visually obvious, or they become too small and we lose the sharpening effect.
For some people, that means that all sharpening must be deferred until the image is at output size, otherwise multiple rounds of sharpening will destroy the image. That's one of the notions this book seeks to challenge: It's certainly true that multiple rounds of sharpening can destroy the image, and it's equally true that the final, critical round of sharpening must be tailored to the output process and applied to the final image pixels.
But I'll be arguing throughout the remainder of this book that optimal sharpening simply cannot be accomplished by a single pass of sharpening at the end of the image preparation process, because the softening that occurs when pixels are turned into print dots is only one of several, often competing factors that must be addressed.
In the next chapter, we'll look at all the different factors we need to take into account when we sharpen.