Tuesday, 17 September 2013

Crisis On Infinite Pixels

Cerebus #207 (June 1996)
Art by Dave Sim & Gerhard


Computer printing "severely challenged" by the task of reproducing Cerebus' tone

I thought I'd provide A MOMENT OF CEREBUS with some explanatory visual material of what the impasse has been about to this point since I know everyone has been waiting patiently for the 30th ANNIVERSARY EDITION OF HIGH SOCIETY and fully restored CEREBUS 16th printing.
Figure 1:
This is using the photocopy function on my printer: my printer which probably cost $60. I took a first printing of the HIGH SOCIETY trade paperback from the Cerebus Archive and -- not adjusting the DENSITY function in any way, just shooting a photocopy of page 207 and enlarging it 268% -- this is what the photocopier came up with.

[I picked page 207 to use as the example because it holds the record for the most ever paid for a Cerebus interior page -- roughly $2700 -- and it was scanned from the original before I gave the original to Heritage Auctions. It seems to me to be worth pointing out what "state of the art" is in printing my "high end" pieces of original artwork. To compare reproduction on this panel, go to HA.com and look at Lot#92285 in Auction #7073, Friday February 22, 2013 ]

Then I took that photocopy and, clipping out the Cerebus head, I then enlarged that already enlarged head 400% (the maximum size my printer/photocopier will enlarge to). The large image is what it came up with. Please note this is a regular "nothing-special" copy of the first printing of HIGH SOCIETY just pulled at random from the 7 copies contained in the Cerebus Archive. If you have a copy of the first printing -- or, in fact, ANY copy of HIGH SOCIETY printed by Preney Print & Litho (which, at the moment, is ALL copies of HIGH SOCIETY) -- this is what page 207 would look like as well if you enlarged this panel from your own copy.
Figure 2:
The inset head on Figure 2 is a photocopy from my same $60 photocopier/printer of the same page, page 207, enlarged the same amount, 268%, from Lebonfon's PROOF of the current HIGH SOCIETY printing.

Now, compare the detail here to the detail on Figure 1. Notice that the dots on Cerebus in Figure 1 are still dots. They've gone fuzzy but then you would expect them to go fuzzy. They're being massively enlarged hundreds of times. But they are still round, black circles. On Figure 2, the best ACCURATE description of them would be "patterned interspaced largely analogous shapes aligned in rows". they are all roughly the same size and they are each confined, roughly, within a comparably-sized area. This is the challenge posed by Cerebus' dot tone screens for today's all-computerized printing
Figure 3:
This is a photocopy from my same $60 photocopier/printer of the same page, page 207, only this is from the "unbound printed copy" which I was sent by Lebonfon after I approved the proof. And I have approved page 207 as seen in Figure 2. Even though there is a disparity between the quality of image between Figure 1 and Figure 2 I always factor in that things that are glaringly apparent to me as a publisher and -- more importantly -- as an artist, are going to either be invisible to or unimportant to the average consumer. However:

Figure 3 is an example of Lebonfon's first attempt at the actual printing. This is what the printed page would have looked like if we had gone straight from the proof stage to printing without having another approval stage (which George wisely insisted on). Or if I had approved this as being an accurate finished representation of Figure 2 based on what I had seen at the PROOF stage, which I just couldn't, in good conscience, do.

Compare the already degraded "dots" on Figure 2 with those on Figure 3. To the extent that you could describe them in Figure 2 as, say, "roughly dot-like" or "dot-ish" in Figure 3 you can't even ACCURATELY describe them as you could those in Figure 2 as "regularly patterned shapes aligned in rows". They are now "irregularly patterned shapes completely dissimilar from one another protruding in each other's directions".

So this is the primary problem: the process of reproducing mechanical tones -- tones applied as "Letratone" adhesive film at the time -- is incredibly difficult with our current technology. It WAS difficult back when Preney was printing the books as well but for different reasons: what Preney was producing was a photographic negative -- basically a super-sharp clear photograph of the art page on a very expensive 100% accurate camera with very bright lights. It was an exacting photographic task to get the image right. But the advantage was: a round dot is a round dot is a round dot. The high-end camera reproduces exactly what it sees in super-sharp definition. No matter how much you enlarge it, it's still a round dot.

That's not true of anything reproduced on computer. The very term "dpi" points to the source or the problem. Dots per inch. You can impress yourself with the fact that you have scanned something at 2400 dpi -- and many do impress themselves with this: HI REZ! HI DEFINITION! -- but a round dot is no longer a round dot. It is actually a grid, a series of squares -- pixels -- that are either white or black. Depending on its size -- how much smaller than an inch it is, what fraction of an inch wide it is (and the dots on Cerebus are all very small fractions of an inch in width) -- that's how many pixels will be required to reproduce each round dot. Common sense tells you that you can't accurately reproduce a circle if you are breaking it down into little black squares. A circle is smooth and round. Squares are right-angled and sharp. That's what you're looking at in Figure 2 and Figure 3. The computer perceives it as: "This area here is 1/500ths of an inch wide. It is black. So it is made up of 8 black pixels. Four of them go here. Two of them stick out to the left. One of them is up on the top. One of them sticks out to the right." And where the 8 black pixels go varies depending on where the dot lands on its 2400 little squares per inch. "On this one, one of them sticks out to the left. Where four of them were on that other dot, here there are only three. Where the fourth one was black there, here it is white. Where one of them stuck out to the right on that other dot, here two of them stick out to the right."
Figure 4:
Illustrates a little more clearly what is going on on Figure 3: This is a 400% enlargement of the same "dots per line" 30% tone that was used on Cerebus, LT17, and what happens when you scan that tone at 1200 dpi. This is what those irregular jagged shapes in Figure 3 actually are: it's called a moire pattern. Because each dot has been "added to" by the dot overlaying it -- some on the top, some on the left side, some on the right side, some on the bottom -- they form new shapes and the new shapes form a pattern which makes it look as if Cerebus is plaid instead of a flat grey. These are all over HIGH SOCIETY. George Gatsis maintains that he's fixed all of these, and certainly that was the case with the PROOF stage (Figure 2). They were still visible to me, but most of them I could live with on the assumption that I have a more practiced eye than most people would have when they looked at the printed book -- assuming that the moire patterns stayed roughly as coarse as they were in Figure 2. Instead, they worsened dramatically on 111 pages out of the 1,000 pages that make up CEREBUS and HIGH SOCIETY. George faulted Lebonfon's quality control and Lebonfon faulted George's digital files.

Basically, I had learned to live with degradation of quality. It's built in. It's the best result you can get with a grid laid overtop of every aspect of the artwork. George had persuaded me that the printing could be much better, while admitting that the problem existed.

[It's worth pointing out that this change from non-computerized to exclusively computerized technology was one of several things that killed Preney Print & Litho: they were unable to make the leap to the next generation of computer printing because, even by 2004 when they were doing the last printing of HIGH SOCIETY, virtually NO printing was being done from photographic negatives or from hand-drawn artwork. CEREBUS was the last. Everything else was digital. The customer supplied the raw material on disk, it was inputted into a computer and made ready for printing and then printed digitally on a computerized press. One of their employees had taken Preney's computerized press system and done a few "work-arounds" -- that was his job as their computer technician, to make their computer system work, so that it was able to print the jobs -- but had, however, unbeknownst to Preney, thus voided the warranty. Which is a pretty common "whoopee cushion" in digital printing I've found out. Your computerized press doesn't actually work so you have to fix it so it does work. But if you fix it, the company that made it says you've altered the equipment so now your warranty is no longer valid if anything major goes wrong -- which it always does. So you end up with a $75,000 inert paperweight and are in need of a new generation $50,000 version of that same press]

Originally, when I started working with them back in 2007, Lebonfon was still capable of working with photographic negatives and making printing plates out of them as Preney had done. And then Lebonfon phased it out. Which was a sensible commercial decision to make, given the hard realities of computer printing having brought any other kind of printing to an abrupt end. Including the technology which could faithfully reproduce Cerebus' tone without creating moire patterns.

It was no longer possible to print on an actual press that was equipped to take that original photographic negative and to make from it a metal printing plate and to use the metal printing plate to produce printed copies. Unless computerized printing evolves further, Cerebus will never again be made up of round dots. Just irregular jagged shapes.

I finally volunteered to pay close to $4,000 to have the worst examples fixed at the printing stage. Not the same as I would have to pay for the entire printing job -- which totals more than $10,000 on each book, $20,000 in total -- but a substantial amount. Patrick Jodoin, my sales rep at Lebonfon, assures me that Lebonfon is capable of printing the books and fixing the 111 substandard pages but always with the qualifier that they can only do that if the digital files don't have the moires on them.

After a months-long delay, I've just seen (September 9th) a set of proofs for 110 of the 111 corrected pages and -- more importantly -- revised first signatures for HIGH SOCIETY (the book is made up of 32-page "signatures", 16 in the case of HIGH SOCIETY) which I have signed and numbered. The quality on the proofs is about the same, in some cases better, in some cases worse, but the signatures are definitely better.

As I say, I do have written assurances from Patrick Jodoin that Lebonfon is capable of printing the books. George and I are now waiting to see a second set of unbound copies of CEREBUS and HIGH SOCIETY to check if the finished printing is closer to the proofs and that the moire patterns are within a "liveable" level of coarseness (i.e. apparent to me but not to the casual reader).

I'm hopeful this is the last update I need to provide on this on-going situation. It's definitely the last one before either a) the books are printed or b) I finish issue #3 of THE STRANGE DEATH OF ALEX RAYMOND (God willing sometime around Halloween).

We'll see when we get there.

Help finance Dave Sim to complete 'The Strange Death Of Alex Raymond' 
by making a monthly donation at Patreon or a one-off Paypal donation.

Originally serialised within the pages of the self-published Glamourpuss #1-26 (April 2008 to July 2012), The Strange Death Of Alex Raymond is an as yet uncompleted work-in-progress in which Dave Sim investigates the history of photorealism in comics and specifically focuses on the work of comic-strip artist Alex Raymond and the circumstances of his death on 6 September 1956 at the wheel of fellow artist Stan Drake's Corvette at the age of 46.


Anonymous said...

I think that if you scale up the size of the printed page in future print editions this problem with the half tone would most likely go away.
Not necessarily IDW Artist Edition size, maybe just 50% up from the current size.
Personally I would rather own larger format version of the Cerebus books to better enjoy Dave's line work.

Jeff Seiler said...

Bet you wish you'd made Cerebus a little WHITE bastard instead of the loveable grey one we all know and despise, eh, Dave?