Part 2

Inking & Screening & Resolution, Oh My!


This is the second installment of Paper to Pixel to Paper Again, a series that explains (in an overly thorough manner) the how-to's of preparing artwork of all stripes for print.

If you haven't done so yet, please go back and read the first installment.

We have only been able to put this together because of the support of our patrons, both subscribers on Patreon, and by one-time donators to our Living the Line Paypal account. Thank you so much for your support! We couldn't do it without your generosity.


Last week we started this series by throwing out some caveats, stating some goals, and peeping at my work space and equipment. This week we'll be laying down the basic knowledge you need on order to know the behind some of the things we'll be discussing and doing.

The past one and a half centuries have seen tremendous changes in the ways images are created, and the ways those same images are reproduced. Since the late 1800s, almost all technical innovation in printing has involved improving the reproduction of what are (misleadingly) called continuous tone images: that is, images that, when viewed in the right circumstances, appear to have smooth gradations of tone and value. If you're viewing a postcard with a reproduced photograph, or looking at a color diagram in a text book, in most viewing distances, these images appear to be smoothly changing values of color. But in reality these images are made up of tiny cells, distributed in an array, that through some miraculous flaw in human vision, work together to create those gradated illusions.

Left: a scanned detail of a Dave Sim commission. Right — an extreme closeup of the resulting print. Notice the array of dots that create the image.

But there are limits to this illusion. Certain people (myself included, at least until I hit forty!) have close vision that's significantly sharper than the mean, and are able to see individual printing dots when they're anything other than the finest pitches. More significantly, the sharpness of human vision increases with a corresponding increase in contrast, meaning that extremely high contrast images (say, black on white) represent much sharper visual acuity than a field of color. Additionally, we see another corresponding increase when we're presented with edges. Lastly, fine information that is near or beyond the fine-ness of the screen itself, or oriented in direction in a way that is not perfectly aligned with the screen, can cause all kinds of unintentional visual oddities.

This is why you will never see a professional publication that has a large chunk of text that is screened and intended to be read. And this is why you should NEVER, NEVER SCREEN LINE ART.

I'm going to belabor this point (who, me?) because it seems to have been forgotten or ignored as screening methods have improved, or as expertise (and money!) have drained from the print fields. Unless you're reproducing in color and intending to show the artist's process as the intent of the print (a la IDW's Artist Series, the Cerebus Archive portfolio series, etc), LINE ART SHOULD NOT BE SCREENED FOR REPRODUCTION.

Above: scans of page 225 of the aborted February 2013 printing of High Society, which, as you can see, was half-toned, as it was supplied to the printer as 8-bit grayscale files. Below is an image from the restored files, produced from the exact same scans as the above images, but treated much differently in the prepress stage. Notice that the dots in the Cerebus image above are strange, gradually changing shapes, moire that is a result of the original screen tone applied to the art being reproduced through a second screening process at the printer.

Imagine, if you will, the simplest of print methods, something along the lines of a stamp. A chunk of potato in which you've carved out an object in relief, maybe your name, carving back the potato from the sections of the images that you wish to remain un-inked, and then pressing it against a flat surface slightly moist with ink, and then finally transferring that ink with light pressure onto another surface.

Your resulting print, any unevenness in the ink aside, is a binary. Either a particular area of the paper is inked, or it is not. On, or off, no in-between. And, really, how would you represent in between? You examine your print some more, consider that you really would like to have a “gray” area of, say, 30 percent to augment your totally black and totally white portions of your image. So you cut some parallel hatching lines into a previously fully inked area of your potato, and you print again.

This is line art in microcosm, line art at the beginning, just beyond scraping lines in the sand with a stick, just beyond taking an old torch and dipping it in bison blood and dragging it along the surface of your cave. Primal, black marks on white, any “gray” an illusion created by finer marks of black on white.


Okay, let's skip ahead at least a millennium, where we arrive at the present day.

No longer satisfied with the speed of your potato print, you're now interested in taking your paper line drawings and reproduce them with all the bells of modern technology. A terrifyingly fast, abominably loud web press, running off a thousand copies of your masterpiece in an hour. Between you and that copy are a good dozen technicians and a veritable space-shuttle level of switches and knobs and little blinky lights. How do you ensure your drawing survives the process? How can you signal to these strange, unknowably distant beings what it is you want out of your print?

(And please don't tell me that your desktop laser printer, or the, ahem, helpful staff at your local copy center, are any more knowable or accessible.)

You need to know how to prepare your files. You need to know what to ask for. And you need to know about resolution.


In order to actually, you know, get to the part of this series where we actually DO something, I'm going to need you to take for granted a few facts. Rest assured I'll come back to them in future installments, and rest assured, I'll be happy to argue with you about them in the comments on the embedded YouTube videos.

When you're preparing color or grayscale images for print, that is, images intended to be reproduced as (not actually) continuous-tone images, the limit of your effective resolution is the screen that these images will pass through. The fineness of a screen is measured in LPI — lines per inch. A printer printing on an extremely coarse surface — a cardboard box, newsprint, some kind of screenprinting application etc — will use a really coarse screen, sometimes as coarse as 40 LPI. Printing on a sheetfed offset press on coated paper, or on a very good one-off digital press on coated paper, the LPI might be as high as 300 LPI.

A good rule of thumb for supplying files that WILL be half-toned is, the maximum effective resolution is twice the line screen resolution. So, if your printer will be screening your final image at 200 LPI, 400 pixels per inch is the highest effective resolution you can supply. Anything above that is pointless, as it's lost in the screen. (This is not the case if you're suppling some elements separately, a in a PDF, where you can have images with different resolutions and color spaces coexisting in the same document. More on this later!)

Conversely, when you're printing WITHOUT a screen — whether that's in black, or using a spot color — your only resolution limit is your vision, and the resolution of the output device, whether that's a laser printer or a plate setter at the printer.

Without further ado, here are the resolutions you should be aiming for for suppling files to your printer —

Color or Grayscale--

as low as 100 PPI in some extreme circumstances, as high as 600 PPI on coated stock with good printing should be dependent on the destination LPI. Remember, it's easier to downscale than upscale! I always scan any color art that's going to leave me permanently (go to a client, etc) at-size at 600 ppi, as a safety measure.

Line art/bitmap--

1200 ppi for laser printers and other digital printers (600 PPI might be acceptable on rough paper if there are no very fine lines or repeating tones present)

1200 or 2400 for web or sheetfed offset with fine lines and tones.

But, fortunately, since we're dealing with line art, just because you're SUPPLYING line art files at that resolution, doesn't mean you need to scan at that resolution!


When we left off prior to those three little dots, we were discussing how you NEVER EVER NEED TO HALF-TONE LINE ART, and how not half-toning allows you to work in a much finer resolution space.

All of the Cerebus Restoration projects have been delivered to the printers as 1-bit (line art) 2400 ppi originals. (In a later installment I'll get into why I arrived at that resolution, and all of the benefits we'd had as a result. But for now, let's just take this fact for granted.)

But just because we're delivering 2400 ppi bitmaps doesn't mean we're scanning at that resolution. This wouldn't be practical or even desirable, for a few different reasons.

Instead we've been scanning at different resolutions (and different formats) depending on the source material, and then upscaling and finally sharpening to arrive at the delivered size. I'll spell all of the how-to of this out in the next installment.

For the most part, there are three sources for materials in the project so far—print materials, photo negatives used to produce the printing plates in the initial run of the book, and original artwork. The print materials and the photo negatives are both at size (or close to the size) of the printed books they will be used to generate, so after much testing, these have been scanned at 1200 pixels per inch. The original art, however, is much larger, and so for the most part these can safely be scanned at 600 pixels per inch and retain all of their detail in the final result.

(How much larger? For most of the run of the book, the active image area of any given printed Cerebus page was around 6" x 9", whereas the corresponding area of the original art was around 10" x 15". For most books the original art was reduced to 60 percent of the original size, with the exception of the latter half of Church & State I through Jaka's Story, which was reduced to 56 percent instead, presumably to try to address any future problems with binding etc in collected editions of the books.)

And why try to skimp on resolution at the scanning stage? There are a few answers to this, but the biggest issue is sheer time. Every time you double your scanned resolution, you're quadrupling the actual time it takes for the scanner to scan the image. (It's not precisely quadruple, as there's a certain amount of setup, warm-up, etc that remains the same no matter how fast the scanner operates).

With the Cerebus restoration project, I was already dealing with a six thousand page problem. Quadrupling your scan time for no good reason isn't ideal.

As to why, exactly, you can safely upscale so much for line art applications when it remains such a general no-no for color reproduction — the very short version is, the color scan contains much more information per pixel than the resulting line art pixel, and that information enables the upscale and attendant bitmap conversion to be very precise to the source document despite the resolution change. More (I seem to be saying this a lot?) in a future installment.


Okay, you've drawn your masterpiece, you've read way too much information about resolution and line art, and now you're impatient to get on with it. Now how do you get all your inky dinky little lines into this here computer box?

Shooting line art used to be a complex process, involving a huge stat camera loaded with orthochromatic film, and an expert operator with a good eye and a ton of experience.

No, uh, not quite that large.

Now, the same goals can be accomplished with a good flatbed scanner and... well, still a lot of skills involved. Hopefully most which can be acquired by reading this series...

Because it was used to both enlarge and shrink artwork, and because every other stage of printing would contribute to the inevitable fill-in of line work, a good stat camera was prized for the resolving power of its lenses. The sharper an optical system is, the more detail is captured from the source, and the less that capture needs to be manipulated to retain that detail. This is still true when we're talking about digital sources.

With that in mind, what follows is a run-down of many of the commercially-available flatbed scanners out there, and what the problems and advantages of each are.

Picking Your Scanner Box Thingie

The Cerebus Original Art Dragnet project has really given me some interesting insight into the wide variety of scanner types and qualities out there. Although the art changed drastically over the years of the book, many of the elements and particular areas of concern from a production standpoint remained the same, meaning that I've had the opportunity to see the same types of screen tones scanned on a huge variety of devices.

If you really, really want to know the optics of flatbed scanners, what's actually happening under the hood and how to really measure the bogus stats that are spewed out by the manufacturers of these products, then visit the fantastic ScanDig and read some of their film scanner reviews. I am indebted to their site and the information they so casually drop in their reviews. And yes, they're selling their own services and there's undoubtedly some self-interest involved in their critical scanner reviews, but they have an incredible bank of knowledge, a professional's eye and a true enthusiast's passion. Good reading.

The Bad News First

There are very few large-format flatbed scanners available, and none that I know of in a middle-range price. And the most affordable ones are, ahem, total crap.

Yes, this means I might insult your scanner. Please prepare yourself.

Mustek ScanExpress A3 2400, or 1200, or any other Mustek

It's hard for me to imagine how a company like Mustek continues to exist in an age where people can read and publish public reviews. They make a demonstrably inferior product that doesn't function as promised, with incredibly buggy drivers, and offer no support or returns. How, you might ask, do they stay in business? Because the prospect of paying $339 for a "large format" flatbed is too tempting for poor cartoonists who just want to scan their damn pages.

Don't do it. Please. Don't do it.

The Mustek uses a special (i.e. super cheap!) kind of scanning sensor that's only capable of seeing with an extremely narrow depth of field. How narrow? If you're scanning a watercolor painting, any slight buckling or indentations in the paper from moisture will be OUT OF FOCUS. Literally the texture of the paper will be out of focus. If you don't load down the top of the scanner with books, the edges or any microscopic lifting at all will result in completely out of focus scans.

Making things worse, the scanner plate is not actually the advertised 11" x 17". Making things even worse, it features a special raised lip that causes any artwork larger than this "almost A3" size to lift off the surface, and thus, to be, you guessed it, completely out of focus.

Add to this motor jitter that causes random lines of the scan to be misaligned with the previous ones, and you have a completely useless paperweight.

UNLESS you only scan almost-11 x 17 artwork that's always completely flat, that has no mechanical tones, you keep some bricks around in your studio to hold the lid down, and you can stomach the idea that this thing will die every few months and need to be replaced, if not returned immediately because they don't power up. Then, hey, do ahead!

From this view, the pages is looking good! Alright, got a good scan. Let's check closer in.

Looking pretty good! Pretty sharp, with no visible artifacts that would indicate any unwanted software sharpening being applied. Maybe this thing really will work out.

Oh, hey, what's happening at the top of this scan??


As you might imagine, having a scan that is overall a little blurry might be okay, given that this is going to end up as line art. Having a scan that is PROGRESSIVELY BLURRIER is next to useless.

Having slagged the Mustek, a useless paperweight, we move on to —

Plustek Opticbook A300

When we started into this project, this was an incredibly tempting piece of hardware. Unlike most other desktop flatbed scanners, it's capable of scanning at incredible speeds—between six and seven seconds per impression. And the large bay and easily movable lid made it pretty tempting as well for the oversize artwork.

So I did the best I could to "test" the scanner remotely. Spoke to multiple sales reps, to their tech staff, got some test scans made of various objects (i.e. portions of dollar bills!) that I could compare on my end to other scanners. I saw that the impression was pretty soft (owing to the type of sensor used by the scanner—speed being its chief virtue, but with a corresponding tradeoff in sharpness), but would be workable with careful post-scan adjustment.

And so we purchased the scanner and had it delivered to Dave Sim in Ontario, and he started the process of scanning the artwork.

Well, my friends, things don't always work in practice like they do in theory. There's some kind of strange...drift to the image capture of the A300 that makes it pretty much useless for anything but its original purpose, bulk scanning of text. It has wildly different exposure levels, seemingly at random, while scanning. This drift happens whether you're scanning with its native utility, or using your own dedicated scanning program. Is this something you can turn off somehow? A function happening in the driver? Unclear. Vuescan, a fantastic scanning program we'll be getting to later, made no difference in the result. I'm inclined to think it's something in the hardware.

Anyway, this drift meant that, before anything could be done with each scan, the exposure had to be adjusted by hand to correct the bizzaro changes from scan to scan, using the known elements of the page to adjust. Over the course of five hundred or so pages, this drove me more than a little crazy, and as other, crisper, scans came in from other sources, I began longing for greener, more color-faithful, pastures...

After working with these scans for the entirety of Church & State I, I implored (begged) (pleaded) (seriously, fax pleaded) with Dave to pretty pretty please purchase this brand new scanner that I'd been able to use over at Bob Chapman's place, a fancy-pants flatbed with great optical sharpness and fantastic color fidelity and...

Ahem. Well. More on that in a moment.

A portion of a randomly underexposed OpticBook scan.

This Opticbook scanner is now the dedicated Aardvark/Vanaheim correspondence scanner, a task it does very well. Really, though, don't pick it up for any graphics purposes. One day we'll live in a world with instant scanning and perfect image fidelity. But we're not there yet.

Which brings us to —

Hey, Can't I Just Scan My Giant Pages in Sections and Stitch Them Together in Photoshop?

Hey, sure, if you need to do this occasionally, go ahead! There are lots of excellent smaller scanners available, including many of the Epson V series, which are great. We do this for the covers and the double-page spreads of the Cerebus books, scanning with a good two inches of overlap per scan and then stitching them together. But if you need to do this more than, say, three images a week, and if you value your time, you should really consider splurging and getting something that will scan it in one shot.

How much do you get paid an hour? Divide the price of your Fantasy Caddillac Scanner by that number. That's the amount of hours of your labor that scanner is worth. It will save you time, it will save you frustration, it will save you that "OH NO I CAN SEE THE STITCHING" moment when you get your masterpiece back from the printer.

That being said, sometimes this is what you have to do. And if you have Photoshop, it's a heck of a lot easier than you think.

Go to File->Automate->Photomerge. And then send me a check for $20.

You're welcome.

Oh hey, how about—

Hey, Can't I Just Put My Pages on A Copystand and Shoot Them With My Digital Camera?

Yes. In the future. When you travel to work by jet pack.

As of now, most DSLR cameras simply don't have the resolution, either in lenses or sensors, to be able to capture at the res you need for line art. They also have various post-sensor antialiasing filters that degrade fine-detail sharpness. Add to this problems with barrel distortion and other lens aberrations when you're shooting a flat piece of artwork without a specially designed lens, and it's just not feasible at the moment, even if you had your dedicated copy stand and $40,000 Hasselblad.

Give it another two decades.

But until then,

Hey, Can't I Just Take My Pages to the Local Office Depot or Fedex Office and Have Them Scan Them?

Well, you sure can, and if you have some Cerebus originals of your own, please do! And then email your 600 ppi scans to cerebusarthunt at gmail dot com. Fun, prizes, etc await.

But for any artwork you have in your own possession, you're better off scanning it on your own equipment.

Every Fedex and every Office Depot is currently using the same leased Xerox all-in-one color units. As one-off color printers, they're fantastic, as long as the person operating them knows which check boxes to un-check etc. But as scanners they're not ideal. They use the same kind of lightning-fast image element as the aforementioned Opticbook, and they automatically compensate with really aggressive wide-band software sharpening that can mess up your reproduction if you use mechanical tones or fine lines in your artwork. This can be turned off, if whomever you're speaking with at the Office Depot is willing to take the time to do so—but it leaves the image itself pretty soft.

A Xerox All-in-One scan, with no sharpening applied.

Overall? Works in a pinch, but less than ideal.

Hey, Can't I Just Use My Prosumer Desktop-All-in-One Thing That I Got For Fifty Bucks and/or From the Costco Dumpster?

If you're using a scanner that wasn't explicitly intended to scan artwork, make sure you test it at the resolution you intend to be scanning at, to see

a. how sharp the impression is, AT THAT RESOLUTION

b. making sure that the scanner software or driver isn't applying software sharpening.

There are a variety of otherwise good manufacturers (LOOKING AT YOU EPSON) who have their scan utilities automatically apply sharpening on their cheaper scanner models. Sometimes this can't be turned off, but you can circumvent it by using a different scan utility. More on this in a later installment.

That said, in the years since I wrote the first draft of this series, there are finally a few reasonable (not fantastic) oversize combo units available that can get you a decent image under the ideal circumstances. In fact, much of Carson's artwork for Strange Death of Alex Raymond was scanned on his Epson Workforce 12" x 18" scanner. Of course, that was under ideal circumstances— completely flat, sturdy artwork. See the embedded video for an actual image comparison between the Epson Workforce and the scanner we discuss below.

The Epson Workforce W-7840 is a fine example of one of these all-in-ones that does well considering the price. But if you're holding out for gold...

So. All this negativity, Sean. What IS ideal? Who DOES make a good scanner?

The Scanner of the Gods: Epson 10000XL (or 11000XL or 12000XL)

I'm not sure what it says about the state of graphics technology or the print industry that the greatest flatbed scanner known to man is more than a decade old now—but there you have it. There are newer scanners. There are more expensive scanners. But for scanning artwork, I don't think there are any better.

Do I need to prove it to you? Or do you trust me at this point?

Let me tell you, I was tempted to skip the last two weeks of writing this, replace those entries with the sentence "Buy Vuescan and an Epson 10000XL." But I think the details matter.

So here's all you need to know —

Ahhhhhh..... Epson 10000XL scan.

Important facts to know before you go out and buy one of these puppies—

a. There are two versions of the 10000XL. One has a plastic lid and can only scan reflective documents, i.e. no transparencies, negative or positive, without the optional transparency adapter (which is actually a replacement lid with a lightsource). The second is EXACTLY THE SAME AS THE FIRST, except it comes with the transparency adapter and Silverfast software! So, if you find one without the transparency adapter, at a really good price, go for it, and pick up the adapter separately. If you don't need to scan transparent materials, well, there's your answer.

b. There is a newer model, the 11000XL, and an even NEWER newer model, the 12000XL. The only difference is they got rid of the Firewire port, and it costs a lot more. If you see a screaming deal on this one, there you go. There's no optical difference between the three that I know of.

c. Don't be scared by the several thousand dollar price tags. If you're patient, and you check Ebay and Amazon frequently, they can be had for less than $1000. Currently there are several listed on eBay. Mine cost $535 plus shipping, after watching Ebay for two weeks...

d. if you DO buy remotely, remember to ask the seller to lock the scanning mechanism before shipping (the lock, which protects the motor from damage and wear, is affixed to where the power supply goes on the back of the scanner)


And that's where we'll have to leave it for now!

I'll be adding new installments to this series weekly, and I'll be updating the master list of posts each time a new one is live. Please leave any feedback you have for us on the comments in the embedded YouTube video!

Please support this work on Patreon, or by making a one-time donation to our Living the Line Paypal account.