Wednesday, 25 June 2014

Restoration Update #1: No Lines Left Behind


Greetings AMOCers, Cerebus restoration backers, aficionados of print esoterica,

Dave and Tim both thought it might be a good idea for me to make a weekly post on the Cerebus restoration work, to give everyone an idea of what's happening with the work, why it's necessary in the first place, and, in the case of the backers, give them an idea of what exactly it is their money is going to.

This is even more timely, on account of the very generous offer/inquiry that Dave received last week from a retailer, who asked, what would $10,000 buy in the restoration effort?

The short answer –

Time. Lots of time for myself and my associate Dr. Mara. Time for scanning source material, time for scanning original artwork in Kitchener, time for file adjustment, time to systematically work through the various reproduction problems involved and get the best method possible for going forward.

I will be able to estimate better after this week's worth of work is done, but as things stand now, $10,000 of time would get us through Cerebus, through High Society, and would start us into Church and State. How far into C + S? Very difficult to say, as things change by the day. Just last week I had three separate incredibly positive breakthroughs that significantly reduced the workload involved for completing the two books. Of course, each of these breakthroughs is in response to set-backs. (A solution to a problem, after all, still involves a problem).

So, what would $10,000 buy us? My current guess – the best versions of Cerebus and High Society ever available. The (not to abuse the word) legacy editions of each of these titles.

Last week I officially started into the restoration work. I began by making a survey of the available materials for High Society and Cerebus both, making evaluations of the condition of the materials, and comparing them with the other alternative materials available to me.

Because the negatives for Cerebus and High Society no longer exist, this is a strange comparison to make, as for the most part I'm comparing inferior versions of superior materials (lower-res or otherwise imperfect scans of the negative) to superior versions of inferior materials (optimized scans from printed material).

When I'd done this first survey, I took some time to prep a few pages from a few different sources, and did some comparisons, both in-file (on the screen) and through "soft proofs" (one-off prints from my laser printer).

This first week's worth of work was almost completely evaluative, and research, attempting to refine a method that's as close as possible to a "one button solution" for prepping pages from different sources of scans. In other words, what I spent my time doing is working up a repeatable, reliable methodology for prepping the pages.

As Dave mentioned on Friday, I've had significant success with this, more than I had imagined would have been possible in such a short period of time.

I've been doing all of my testing with my best Charles Dana Gibson book, mainly because his work has some of the finest linework I've ever seen, the kind of lines that break up very easily without the right treatment – and because the samples I've been working with have radiating lines, another thing that makes any flaws in the scanning or the output very obvious.

Here's a raw 1200 ppi grayscale scan from the book. This represents a 1 inch by 1 inch area of printed material, probably 3rd generation, at best. You can see a little line breakup has happened at some step in the journey to printed material, possibly due to bad exposure when shooting the original artwork, or bad exposure shooting this particular book, or possibly the lines were gray-ish and watery in the original, and this is the printed consequence.

Regardless, our goal is to get this to look exactly like itself in print, and to output a 1-bit bitmap file that captures all of the information that is currently present on the page.

So why exactly isn't there a "make into line art" button?

This extreme closeup might make it clearer. Look at the edge of each of these "lines". You can see the gray pixels that make up each of the edges here. Even though your brain is perfectly capable of interpreting these as lines, the computer doesn't know the difference between those gray pixels and any other gray pixels of the same density. This includes gray pixels in areas that are totally inconsequential to us, including "noise" on the paper. In other words, if you just straight convert the file now, the edges of the lines (and many of the lines themselves) would simply disappear.

Here's what that would look like.

So that's obviously not going to work. This is the point where many people prepping files reach for the burn and dodge tools. In other words, they're going to "darken" (thicken) the thin lines and "lighten" (thin-en (?) ( ^_^ ) the dense/dark areas to avoid this problem.

But to me, this is unacceptable. There has to be a way to preserve the line densities of the original artwork and not have line breakup.

And indeed there is.

Step one, in the case of really fine lines, is to work in a higher resolution space. Step two? Sharpening.

Conventional wisdom on this involves Photoshop's Unsharp Mask, but Smart Sharpen involves a much wider range of control of the adjustment, which is especially necessary with such fine lines, and when the material being scanned is on paper so close in contrast to the ink darkness.

After several hours of testing and printing last week, using a variety of different filters and different source materials, I now have what I consider to be as close to a one-click solution as possible. I have worked up a recipe of settings using Photoshop's Smart Sharpen filter that are, along with levels adjustments and a final threshold adjustment, are giving me in-file results that are identical with the printed source.

Over the next few days I'll be using this method to start prepping the pages for the replacement signatures for the Cerebus book. Once I'm confident with the method and I have it down to a formula, I'll be joined by my colleague Dr. Mara, who will be batch scanning more source material for some pages, and then working with me to prep the pages. The files will be delivered to Lebonfon by July 11th, and printed later that week.

After that? It depends on where we are in terms of the Kickstarter funds.

The majority of those funds will be going towards printing bills and fulfilment of pledges, but the remainder will be used to finance this work. I billed Aardvark-Vanaheim $506 for last week's work, and will continue to bill weekly until Dave lets me know that we're close to the funds being out.

I'm doing my best to make this work economical, and this inevitably means that the costs are upfront to ensure the best and most economical result in the end. Better for me to work 20 hours to produce a replicable formula then charge in head-long into work that will need to be redone. All of us here are in for the long haul here. The best reproduction possible. Or, to take a page from my previous vocation, No Lines Left Behind.

Next week I'll try not to be so dry. I'll share some direct comparisons between different methods of reproducing a single panel of Cerebus artwork.

And in the meantime, if anyone feels like contributing to the effort, let Dave know! Or keep your eye out for the next Kickstarter... 

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


Eddie said...

I didn't find that dry at all. Especially as I've been trying to figure out how to print scanned text and images at a sharper output myself. I wonder if you might want to keep some of the techniques and procedures you're using (pioneering?) secret, for proprietary reasons, since this might be new territory you're charting?

Jeff Seiler said...

Sean, this entry was the most approachable of all of them so far. I have to admit that my eyes glazed over while reading many of the previous entries, wherein you and George debated the various techniques. The above sort of thing, however, is a very good way of keeping us up to date on the procedure/s. Please keep it up, especially with the comparative images. And,


Sean Michael Robinson said...

Hey Eddie,

I have no idea if someone else has this exact formula or not, but a glance at most books with poorly reproduced lineart says "no." :) As far as being secretive about it or something, I'm not sure. I'll give it some thought. My first inclination is that, if other people can benefit from it, why not share?


Thanks! I don't blame you for glazing over. I wasn't trying to communicate to a general audience before, only trying to be persuasive in a fairly technical manner. Many many more illustrations to follow, I promise.

Iestyn Pettigrew said...

I thought it was great to hear this and I personally think you're right that it is good to share this information in case others can use it. There are so many old black and white comics that should come back to print and you may help this happen by sharing this.

You'll still be able to show your skills and win work that way if need be, but I feel like you're not really doing this to drum up business!!

Maybe Dave could add all these technical scanning things to the Cerebus Guide to Self-Publishing and publish it for the digital era with webcomics and tumblr and blogs as well!!

Dave Kopperman said...

These are looking good, Sean. I second the desire for you to share your adjustments (I presume you're setting it up as an action?).

How does it handle remaining pencil lines and other non-ink artifacts?

Sean Michael Robinson said...


Very well! Give me a little time and I'll share specifics in one way or another. So far haven't made an action yet, as there are a few added operations depending on the source material... That's the goal, tho :)

Chris Judge said...

I'd like some insight into your methods too, I draw a lot of tiny little lines and have mostly been able to preserve them for print, but any little bit helps, yeah?