Cerebus Restoration Milestones- Second Birthdays

Sean Michael Robinson:

Greetings, Friends of Cerebus!

It's been a while since I've updated AMOC, or indeed, anyone but Dave and Sandeep, on the state of the Cerebus restoration. It's now been a little over two years since I started on this project, which has so far restored 2,300 or so pages and has taken literally thousands of wo/man hours.

As the time between updates increases, I have more and more to write about, which makes the actual writing more and more difficult. So forgive me the brief summary I'm about to indulge in.

What's Been Happening/What's the Status of My Favorite Cerebus Book?

READS- status: Complete

The work on Reads, including my long and involved end notes/essay, is done! The book is in production as I type this, mid-job at Marquis, our new printer. They've been fantastic to work with so far, very thorough and detail-oriented. As a for instance—it's not every prepress department that will kick back a cover because the writer's names aren't centered on the spine to the millimeter. Reads will be completed on the 5th, and will be shipping to Diamond shortly after. This will in all likelihood be the best-looking Cerebus trade yet.

Church & State II- status: Restoration Complete, awaiting essay and back matter

Of all of the books I've worked on so far, this one and Volume One (see below) will have benefited the most from the work we've done. Part of it is the sheer range of techniques that Dave and Gerhard tried in the book, and how much better those techniques can translate now to the printed page. The snowstorm. The moon. The stars and other celestial events. The Scuwws. All textures dependent on tiny flecks of texture or dense dark areas that easily fill in on photography and when printed on porous paper. When this book sees print (i.e. when there's demand for another printing!) it's going to blow some minds.

I haven't finished the back matter for this book yet as what shows up there is somewhat dependent on paper roll size, i.e. the size of the signatures we can get for the book. We're printing Reads with Marquis, an Ontario printer, on a paper that's not typical for them, that they have to special order for our jobs. Because of this, it's expensive for them to have a variety of paper roll sizes. Reads was printed with large 32-page signatures only. Since the current size of C & S II is 640 pages (20 x 32 pg signatures), this means that any pages we add will have to be a new complete signature. 16 pages of back matter, on another size paper roll they have to special order? 8 pages, and have them hack off half the signature and pay for it as waste? I'm not sure of the right decision for this, and it'll likely change over time as the paper availability and use does or doesn't change. Hence, not yet finishing the back matter yet, since it might need to get cut for publication. I do, however, have a draft of the essay.

Going Home- status: restoration begun, halfway through cleanup

Surprise! Going Home ended up being the next book we'll be printing. I'll have several posts dedicated to this starting next week, so let's skip forward for a moment to-

Cerebus Volume One- status: Restoration Complete, awaiting revisions, layout and essay

Those of you who have been following along for the two years of this restoration project will remember that it began when Dave was having problems with the sixteenth printing of Cerebus Volume One, and the eleventh printing of High Society. Unbound copies of the book had been printed but had mysterious moire across the images, and the line art seemed soft and indistinct. This dispute—what had happened and whether the printer was responsible—went on for almost a year before I ungraciously butted in, getting into a protracted online discussion of prepress and print techniques that finally turned into a general agreement for me and my colleague Mara to produce four replacement signatures for the Cerebus volume, restoring the pages to the best of our abilities at the time and replacing the pages most effected by the moire.

Ever since then, each book we've worked on, the plan was to one day return to the first volume, and complete the work we started.

And now it's done! It has been without a doubt one of the most labor-intensive projects I've ever completed. I say this as someone who's co-written and illustrated a book, written engineered and produced a half-dozen albums from the ground up, and is currently raising a small child.

Why so intensive? And why did this work have to be done in the first place?

Well, it turns out, there has never been an edition of Cerebus Volume One printed from the original negative.

In fact, the majority of the Swords editions were also printed from negative "dupes" of the original photography, most likely made from printed copies of the original issues. 

How do I know? From studying the original monthly issues.

There are three major types of source material used for the Cerebus Volume One restoration. The first, and most frequent in appearance, are the original monthly books, specifically, the Dave Sim File Copies sent to me by Peter Dixon of Paradise Comics. 

Below is an image from my recent restoration work, scanned from the DSFC of the original monthly issue #4. Next to it is a scan of my 6th printing copy of the book. As near as I can tell, this issue was only printed once from the original negative—when it was originally published. Every other time has been from a replacement negative made from a print copy of the book. Of the 25 issues in the first volume, only 8 retained their negatives through the Swords of Cerebus and Cerebus Volume One printings. 

Even if you have an original printing of issue #4, you'll find that the page in question appears much more filled-in and dense than the artwork above. That's because, during the course of the work on this volume, I figured out a technique of reversing dot gain, that is, shrinking every area of a finished page uniformly in order to reverse the amount that the ink had spread during the printing process originally. 

This might sound a bit crazy. It certainly would take a crazy amount of time to explain it and explain how I tested it to make sure it really works the way I thought it works. The very short version—dot gain is the amount of spread that, inevitably, occurs when you print. Depending on the surface (substrate) you're printing on, the print process, the ink formula, etc, you get more or less of this spread. On a web press it's not even uniform—different parts of the form, and even sometimes across each impression, have varying amounts of dot gain. But mostly, per page, it's fairly uniform, and when it's not uniform, it's usually visible. 

The visual effect of this spread isn't linear—i.e., areas that appear dark/dense fill in quicker, because the ink is welling in from all sides. But the actual process, i.e. if you measure the amount of spread of the ink, is linear. So, if you had some way of shrinking all areas of black on your page uniformly, and you had a definitive metric that you could use to calibrate your correction, like, let's say, a thirty percent dot tone on every page, then you could conceivably reverse dot gain and still preserve all of your detail.

And so that's what I did. The trick only works because I'm working in an outrageously high resolution space—2400 ppi. I finished each page, restored and made the most pristine copy/capture of every detail in the original printing, cleaned up all noise in the newsprint, and then flattened it to bitmap. And then applied a very low radius gaussian blur to the entire image. Brought up the levels command, lowered the mids/exposure until the tone I was using to calibrate reached the desired density, then after the levels adjustment, sharpened the entire thing at the same radius that the blur was at initially. The result? All areas of black uniformly shrunk, all detail preserved. In a case where there was variation in visible gain across the page, and that gain happened to line up with the tiers/panels of the page, I adjusted the tiers separately to the same end.

The end result is the reversal of any dot gain in the original printing that didn't completely fill in an area of darkness. The result is, frankly, better than the original printings, despite now being one more generation removed. And the resulting book is going to blow the earlier printings of the collected volume out of the water.

(There are other ways to open up line work, but I've been reluctant to use any of these "spot-applied" methods, as it changes the overall tonal balance of the page. As a pen and ink illustrator, I'm sensitized to tone in black and white work, how those tones and textures interact with each other. If you "open up" one area of dense crosshatching from, say, 70 percent to 40 percent but leaving surrounding tones alone, what have you done to the balance of that area of the page? If you "open up" the clogged tone of a figure but leave the remainder of the page filled-in, how have you affected the balance? Therefore, I've restricted myself in this work to linear processes that act evenly across an entire selection. It's not enough that every line drawn remains, but that they remain the same size in relation to the surrounding lines and tones as well).

The second major source of artwork was a combination of these file copies with the only digital scans ever made from the original negative, made way back in 2004 (?) or so by Lebonfon. This neg scan was a copydot scan, and had a host of problems of its own, and it's rather deserving of its own post how I was able to wrangle it to work with print copies. What follows is the short version.

The copydot scans are 1-bit direct-to-bitmap scans of the original negative (or, when the negative had been replaced, the dupe negative). The exposure had been dramatically spiked when they were scanned, which caused much of the fine-line information, fine tone, and gray brush strokes to blow out, and made each page much lighter than it was originally drawn. These are the scans that were the basis of most of the sixteenth printing work. But, being direct scans of the original negative, the dot tone was pristine, with none of the junkiness of tone you see from dot tone printed on pulpy newsprint. It also had information in the dark areas that had filled in in the initial printings. So if only there was some way to combine these scans with print copy restorations to get the best of each type, without having a mismatch in exposure...

Well, with the "de-dot-gaining" technique I discussed above, I could do just that. Converting the bitmap to grayscale and bringing it to the right resolution, a very very narrow radius gaussian blur, and then I could re-expose the page using the tone as reference, bringing it to where it had been previously. Then completely restoring the Dave Sim File Copy version of the page, "de-dot-gaining" it to the same exposure, and placing the two together via Auto-Align Layers function in Photoshop. And finally, figuring out which one had the most information I want to keep, and erasing or otherwise eliminating the rest.

Here's a peep at one of the pages restored using this technique. Above is the blown-out scan from the negative, and below is the combined page, restored from a combination of the negative and newsprint. 

Labor-intensive? Check. A little crazy? Check. Pretty? Check.

The last source of artwork?

Of the 538 (?) pages of art in the book, a full 94 pages of those are now sourced from direct scans of original artwork—almost double that over the previous printing. It's been very slow lately, but the Art Dragnet is still pulling in pages. 

As always, the original artwork is in a class by itself. Even though the tone is a wreck on virtually every page I've seen, the amount of detail that can be found is really a revelation, almost to a page, as detail that would have previously been too fine to be captured by stat camera or would have gained up and filled in after being printed is able to be revealed for the first time. (Including, sometimes, unintentional "detail" that needs to be drawn out—most frequently, razor blade white lines from the edge of where the tone was cut.)

If the image above doesn't say it loudly enough, let me be more explicit—We need your original art scans! We need your friend's original art scans! We need your local comic shop's original art scans! We need your wacky neighbor who has hoarded two hundred pages in his closet and forgotten about them's original art scans! Seriously, where are these pages? I know they're out there somewhere... Any leads? Email me at cerebusarthunt at gmail.

So, that's the state of Volume One. With a book this long and a process this, uh, complex, there are bound to be revisions once I've had a bit of time away from it, and had a chance to make oversized prints of the work so far. (In fact, I'm seeing a bit of additional cleanup just putting together this demonstration here...) But for the most part, the labor is done.

Now it's time for your part.

What, exactly, would constitute for you, Cerebus Reader, as the Best Possible Version of Cerebus Volume One? We're already enlarging the image area (104 percent of previous volumes), and presenting better paper/printing and Smythsewn binding so that the book opens flat. But is there anything else that should be done to this volume when it finally goes to print? 

Dave and I had discussed the possibility of adding the Swords of Cerebus intros back into the book, but at the rear this time, as annotations, so that they don't interfere with the readability of the book a a whole. Does this seem like a good idea? Bad idea? Keep the banners? Keep the layout? Any other changes you can recommend?

If you've made it this far, I commend you for your endurance. I look forward to your comment.