Thursday, 3 July 2014

Restoration Update #2: tick tick tick


When I talked to Dave about working with Cerebus pages at a higher resolution, the first thing he said to me was, something along the lines of, "when you press down a bump in the carpet it comes up somewhere else."

Well, we keep pressing down the bumps.

My newer, faster computer arrived on Tuesday, and Dr. Mara and I started into the project in earnest.

Some of the roadblocks we're hitting have everything to do with working at such a high resolution. Although the resulting files are actually pretty small, being bitmaps, the in-process grayscale files are massive. Just for some perspective on this-- my former computer, now being used by Dr. Mara in my studio, can run upwards of 50 channels of high-resolution, 24 bit depth audio, all with plug-ins, without blinking. Opening a single one of these massive files and attempting any kind of manipulation brings it to a stand-still.

What this means in practical terms is, for now, I'm doing the majority of the adjustment myself on my new computer, and until it's financially possible to purchase a second high-end workstation, that's the way it'll stay.

The good news is, I've set up several of the processor-intensive actions as “actions” in Photoshop, meaning that I can press the “go” button and it starts the work by itself. For instance, I'm typing this now to you all on a laptop right now while Photoshop works away at the workstation.

Dr. Mara, meanwhile, is scanning pages from the Swords volumes that Dave sent me last month, to replace the negative scans we were supplied.

This is the real major roadblock, one George told me about and one that me and Mara are experiencing for ourselves now. Basically, everything is sourced from different places, everything is imperfect in some way. In short, everything is a mess.

I have two separate sources for the destroyed negatives right now, and have had sporadic access to a third source.

The first source of negatives have the deadly combination of blurriness and excessive sharpness on the edges. More critically than that, they have scan “hitches,” line by line errors, possibly caused by some kind of mechanical error in the scanner. On the finer line, these hitches transform them into broken-line dots. They also have periodic vertical hitches as well, missing several lines of vertical information in the file. The latter is unlikely to be very visible in print, but the former gives the appearance of a low-res scan, with visible stair-stepping in lines that intersect the error at the right/wrong angle. (What it actually looks like is some kind of ineffective digital interpolation with automated sharpening, but there's no way to really know without having the scanner here in front of me)

The second source of negatives are lower-res than I'd like, and have a severe “negative shadow,”meaning that the negative was mounted in such a way as for the positive areas to cast a shadow on the scanner bed during scanning. However, unlike the first batch, this can be corrected with an automated process involving levels adjustment, blurring and upscaling.

The third source is Lebonfon's “copydot” scans, which are 1-bit 1200 ppi scans from the negative. I've only seen a few pages of these, as they haven't given me access to the whole batch, but they're useless to us for a whole other set of reasons. Although they're very sharp, the 1-bit 1200 ppi sampling has blown out all the fine line areas into broken lines, or made the finer lines disappear entirely.

A small area of Lebonfon's 1200 ppi copydot scan of page 186 of High Society
The same area of the lower-res grayscale negative scan.
Notice the presence of the blown-out lines.
This is the problem I alluded to last week, of having to choose between flawed scans of better sources, or better scans of flawed sources.

The best sources are, no surprise, the rare cases where we have high-resolution scans of original artwork. Even the lower-res scans are incredibly useful, once I've applied Currently "Secret" Procedure #2 to them. (It's a little surreal to see smooth hi-res line art arise from a low-res color scan.)

It's really amazing to sit down and adjust a page from original art and compare a printout to the original books, to see how much information really was left in the dark room. Between the bad/underexposed photography of the early books, to the inevitable dot gain/line fill-in of each step of the former photographic processes, a tremendous amount of visual information just filled in, became a mass of black rather than the fine lines that constructed it.


But but but.

The original pages need more attention than any other potential source.

First off, unlike scans of the negative or printed matter, the original pages have never been contrast adjusted. My "one button" process is handling this well, but pages that have a significant amount of pencil or white-out/reverse drawing with white need more individual time/contrast adjustment of those specific areas before moving on to the final stages.

Additionally, all of the older pages suffer from tone shrinkage. All that Cerebus dot-tone is essentially emulsion on acetate with a sticky binder to keep it on the page. Over time the acetate shrinks, leaving gaps between the tone and the outer edge of the area it was applied.

Well, Dave asked me, how about fixing it?

It's relatively simple to clone the tone from adjacent areas and move it into the gaps. But like anything, it takes time. How much time? For these early pages, somewhere between 5 and 35 minutes, depending on the amount of tone on the page. Conceivably much longer than that on, say, the tone-heavy Fall and the River. This equation might easily change, however, when we get to the later books, which were of course, better photographed, and being comparatively younger, are less likely to have significant shrinkage.

A newly fine-lined Jaka examines Cerebus' tone shrinkage. 
From Cerebus issue 6, sourced from original art.

So, is it worth it? How much is it worth?

That is for Dave and the rest of you to decide.

Meanwhile, I'm excited and thrilled by how good the finished pages look so far, and very hopeful that in the near future, when the options narrow and the materials improve, that the way forward will only get easier.

Next week we'll be delivering the replacement signatures to Lebonfon, and we'll have some news on the original art front... In the meantime, let me know what you'd like to see here, what's interesting to you, and what's dull as dirt. More absurdly close-up pictures of tone? More action? More drawings of cats sleeping on unbound copies of Cerebus and High Society?

(drawn by one of my cartooning students, the awesome Sofia Angulo, during a “job shadow” last month. I'm not sure what she learned, other than the fact that freelancers do approximately 30 things at the same time...)

Sean Michael Robinson can be found online at Living The Line.


Margaret said...

stupid question: will all of this hard work show up on the newsprint used to print the books? I can understand if the 'legacy' edition of the phonebooks is a high quality paper that will hold all these super fine lines, but it seems like the newsprint wouldn't be able to. . . Unless it can? I don't know enough about the printing limitations of different types/weights of paper.

I just hope in the future that we can get a phonebook print that has all this fine lines and hard work shown, as I think it'll show the original art in a new light. That is one thing I've seen with Dave and Gerhard's original art, so much more then what is in the phonebook. . .

Sean Michael Robinson said...

Hey Margaret,

Not a stupid question at all!

First off, the new printing we're working towards now, four replacement signatures for the 16th Cerebus printing, is not on newsprint. Newsprint doesn't have the same competitive price advantage it once had, as the paper market has changed drastically over the past two decades. This is rough uncoated stock maybe a little heavier than your average copy paper, and it's denser visually (less peek-through from the adjacent page).

So, yes, newsprint is notoriously porous, and you get a lot of ink welling/"dot gain". But it's still capable of carrying lots of fine-line information. Take a look at how fine the lines are in the later Cerebus volumes, and even parts of High Society and a few places in the early issues, and you'll be surprised. (When I was testing the methodology two weeks ago, I was using a newsprint scan of a Silverspoon page that had unbelievably fine hatching lines on his face, with the thought that if the method worked on a newsprint scan of those tiny almost not-visible lines, then we were truly in business.)

Anyway, the short answer is yes, newsprint can hold much more information than you think, but we're not printing to newsprint, and it's unlikely future volumes would be on newsprint either. I'm also endeavoring to make all of this work something that will never have to be done my anyone else ever again, hence the (most people would say) overkill in the areas of resolution and attention to detail.

Lastly, I'll go into this more next week, but it's worth remembering that the loss of detail and fill-in comes from many sources not related to the paper, including under-exposure of the original photography, not shooting with fine-grain negative film, clumpy emulsion on the negative, the multiple generations involved in production of the page (neg to plate), etc. Once the major hurdles have been solved, computer-to-plate has the potential to be a better, cleaner printing process than the traditional methods.

Sean Michael Robinson said...

Another thought-- I wanted to point out that the later books is also informing the work on the earliest, specifically in the area of resolution, but in other areas as well.

Unknown said...

I for one find this all fascinating, and look forward to reading more. I'd also like to thank you for all the hard (and tedious) work you are doing on this. Cheers!

Jason Winter said...

I appreciate all the work you're putting into these new editions, the updates are very informative and enjoyable. I'm looking forward to reading more.

David Birdsong said...

I love hearing about all these fine details so please keep sharing.

Sean Michael Robinson said...

Thanks everybody!

Hey Unknown,

Nothing tedious about it! I'm enjoying myself-- I haven't had a challenge like this in a long time, something that's both quasi-artistic and quasi-scientific. (Mara keeps helping with the last bit, asking about "inter-rater reliability" in reference to tone density... :) )