A book report: stochastic screen pitch

This is post four in a series about my experiences in publishing a book. The series starts here.

There’s another decision that I have to make about the book. Do I want it printed with 10 micron or 20 micron plates? That probably doesn’t mean much to you. But, if you’ve got some inkjet experience, you know more about this than it appears at first blush; it’s just that the printing terminology is unfamiliar to you.

Let’s back up and talk about halftoning. For the most part, printing presses can’t lay down more or less ink at one microscopic place on the page; your choice is the presence of absence of one or more of the inks that are loaded in the press. Going from an image in which each pixel is specified by one or more multi-bit values to one where the presence or absence of each ink is specified by a single bit (one means put ink down there, zero means don’t put any ink down) is called halftoning.

Before computers and printing presses began their courtship, halftoning was performed photographically, buy shooting the image to be halftoned through filters and screens onto lithographic film, which, when developed, was either pure black, or purely transparent. Those films were used to make the printing plates. Almost always, the screens had a regular pattern to them. Screens meant for newspaper printing were coarser than screen meant for coffee table books. The coarseness of the screen was specified in lines/inch or lines/cm, even though there weren’t any lines in a screen; the lines referred to the pitch of the dots. To make it even more confusing to the layperson, the denominator was almost always omitted when talking about screen pitch. Your printer would say to you, “Is a 133 line screen OK?” and you were supposed to know that he meant 133 dots per inch. By the way, the screens used in newspaper printing are so coarse that you can see what’s going on easily. Take out the Sunday comics page, and look closely if you’re young, and through a magnifying glass if you’re not, and you’ll see that there are four screens, one for each color, and that they run at four different angles, forming rosettes on the paper whose shape and color vary with the color being printed.

When computers started to do halftoning, at first they more-or-less emulated the screening operation formerly done with cameras. It didn’t take long for the computer people to come up with a better way. The name for the improved halftoning algorithm is different depending on the context. Imaging engineers call it “error diffusion with blue noise dither”, a phrase that doesn’t flow trippingly off the tongue for most people. If you’re in the printing biz, you call it “stochastic screening”, which ignores the error diffusion part and focuses on the noise, and might be a bit confusing to those outside the business since there’s no screen involved. For the rest of this post, I’ll follow that convention and talk about screens, even though, strictly speaking, three aren’t any.

If you’re an inkjet printer, you use stochastic screening all the time but probably don’t think about it at all; it’s built into your printer driver. While an inkjet printer is a bit more flexible than a printing press, since it can vary the drop size and lay down overlapping drops, it still needs halftoning, and the driver does that.

Now we have to deal with a topic that confuses most photographers when they start to use inkjet printers: resolution metrics. The pixels in your image in Photoshop or Lightroom have many bits per color plane; that makes them continuous tone, or, in the jargon of printing contone, images. When those images are mapped to a printed page, their pitch is usually specified in pixels per inch, abbreviated ppi. Your printer has resolution limits, too. In order to keep from confusing them with the pixel pitch, the resolution of the printer is usually specified in dots per inch, or dpi.

As an example, the Epson 4900, which I know and – when it’s not clogging – love, has several dot resolutions: 2880 x 1440 dpi, 1440 x 1440 dpi, 1440 x 720 dpi, 720 x 720 dpi. You pick the one you want in the driver dialog box.

However similar the concepts and algorithms of the inkjet and the press world are, the terminology confounds. In the print world, a 10 micron stochastic screen means that the dot centers are that far apart. That translates to 100 dots per millimeter, of 2540 dots per inch. If it were an inkjet printers, we’d say that the resolution of a 10 micron stochastic screen is 2540×2540 dpi, which is a little worse than the best resolution of the Epson 4900 in one dimension, and a lot better in the other. A 20 micron stochastic screen translates to 1270×1270 dpi.

On my Epson 4900, switching between 2880×1440 and 1440×1440 makes no material difference in the colors. The Hemlock folks say that that is not the case with their presses. A 10 micron screen yields deeper, richer colors than a 20 micron screen. There’s another wrinkle, too. They have proofers that are capable of proofing 20 micron screened images, but they can’t proof 10-micron screened images; for those, you have to look at the 10-micron proofs and imagine what the image will look like coming off the press.

It’s a leap of faith, but I’m currently favoring a 10 micron screen.