Tuesday, 13 June 2017

Paper to Pixel to Paper Again: Part 19

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 9 | 10 | 11 | 12 | 13 | 14 | 15
16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29

A guide to creating the best looking line art in print in the new digital print world

Part 19
Working With Multi-generation Sources


This is the nineteenth installment of Paper to Pixel to Paper Again, a series that explains (in an overly thorough manner) the how-to's of preparing line art (and later in the series, color art!) for print.

And as always, if you have any questions, please let me know in the comments!

Alright, everyone, this is the big one, the reason (probably) that I'm writing this thing in the first place, the place where there's the biggest gulf between the general "received wisdom" and what's actually possible in print—working with multiple-generation materials. That is, restoring work from print copies of that work.

For the examples in this section, I'm going to be using the first issue of Cerebus, since it's a. very well-known to the AMOC crowd, and b. looked so terrible over the years. If you have a copy of the newly-restored Cerebus Volume One (out from Diamond in January 2017) and an older  copy of the book as well, you can follow along with the books.

But first—some background.

All Else Being Equal, Closest to the Source is Best

Just because you can produce working versions of books from multi-generational materials, doesn't mean you should. All things being equal, if you have access to (or can otherwise dig up!) earlier generations of materials—be that original artwork, production negatives, blue-line proofs of a book (which don't exhibit dot gain), or versions of the same material printed in a better manner on paper that exhibits less gain (tear sheets, for instance), then do so! It is almost always better to fix your image at the source than to try to "fix" something after the fact.

That being said, materials disintegrate. Negatives are thrown away or recycled. Tone shrinks. And sometimes what scans you have of an earlier generation might be inadequate compared to what you could have done were the actual materials still on hand. Hence, this series.

Different Editions of Materials Will Have Wildly Different Qualities

This goes along with the maxim above quite neatly. If you're restoring a work that's been printed multiple times, different copies, and especially different printings, will most likely have very different qualities, and the qualities you'll be looking for are different depending on your intentions with the restoration you'll be doing. Do you want as little to fuss with as possible, as little manual cleanup as you'll have to do? Then you might look for a printing of the material that has richer black than others. Want as much detail as there is to get? Then you'll most likely want to find the opposite—a lighter printing that has less dot-gain overall, and has a corresponding level of increase to the amount of detail in the darker, finer, denser areas of crosshatching or tone.

For my work, it's only been the latter, looking for printings that exhibit the finest details with minimum gain. In the case of, for instance, the High Society material, the Bi-Weekly reprints often have much more fine detail (and much weaker black) than the corresponding monthly issues, and the collections almost uniformly have much less detail than even the monthly issues. If you're serious about getting the best possible result, then checking a range of printings side-by-side is the way forward.

More caveats along the way. But now it's time to dive on in!

Dr. Mara scans the first issue of Cerebus waaaay back in 2014. My how time flies!

As I said in the above, different printings of the same materials can be wildly different from each other, even when the printer was working from the same negative. The first issue of Cerebus (and indeed, the majority of the first 25 issues) are a special case, in that the original negatives appeared to be discarded after their initial use. So barring the original art showing up for the first issue, working from the first printing print copies are the best that it gets.

As with the negatives, we're going to scan at 1200 ppi. Unlike the negatives, this time we'll scan in color, as it will enable us the most flexibility possible for manipulating our scan after the fact.

Here's the raw scan of the first page. Special features—Deni Loubert's hand-written notation in the corner, indicating...what, exactly? I'm not sure. This copy (and the other copies I have in San Diego of the first 50 issues) were the office "file copies," kept around to refer to when reprinting an issue, making an ad, etc, or any other time the issue needed to be referenced. So oftentimes they have mysterious markings, post-it-notes, etc in them.

We're going to write a script for working with newsprint, but unlike our other scripts, this one will require some work on the back end to get the most out of it. And first, we'll need to adjust the, erm, lean that this scan currently has.

When I restored this book, I did the rotation (and making the image grayscale) in advance for all of the pages, using Lightroom, as as I indicated in an earlier installment. But for now, we're going to do this in Photoshop.

Bring up the Layers menu, double-click the layer and then hit Enter to unlock it, and then hit Control-T to bring up the Free Transform feature. Now align the border boxes with the grid, and when you're done, hit Enter to complete the transformation.

Okay, now that our page is straightened, we're going to take a look and really see what our blacks look like up close, and if any particular color channel in our scan is better than another. 

Zoom in on the portrait of Young Dave Sim in the second panel. (What's that? You never noticed Dave's appearance in Cerebus,seventeen years before Minds? Well, he's also in issue 4. See for yourself!) 

Up close, we can see that, although the blacks are far from rich in color, there's not quite the snowstorm in the solid blacks that I was expecting. So this will be a bit easier than it appears.

I'm now going to bring up the Channels control and flip through the channels to find if any particular one has a stronger image (or less noise) than the others. (If you were doing this in Lightroom, the procedure would be different, but the principle the same).

I flip around on the Channels dialogue, selecting each one in turn and deleting the worst-looking ones until I'm left with the Yellow channel as the sharpest impression. Now I change the color mode of the document to Grayscale.

Here's the resulting image.

Okay, now we're going to bring up the Levels command (Control-L) and try to knock out our paper noise and bring the blacks up to a reasonable level of blackness.

Keeping my eye on the histogram, I move the Blacks point (far left arrow) to the first peak, shaving off the empty area of the histogram. Then I move the White point to the left in a similar fashion, while watching the finest areas of detail to make sure they're not disappearing or otherwise being modified. Once you have these in place, you can move the Mid point (the Gamma correction/exposure/gray arrow) to see if you can either a. get a slightly richer black without losing any detail, or b. bring any hidden detail out in your dense dark areas. If your scan is good, either is unlikely, but give it a shot anyway.

And now our page is ready for the next stage! Save (or save-as) and then start your script.

(Why can't we script this stage as well, like we did with the original art? Because there's so much variation in the blackness from page to page, especially if an issue was printed on a web press, and on newsprint, both of which are true of every issues of Cerebus. You can script this, but for the best result, do this stage manually, and then script the rest. Or, if you prefer, break your script into two stages, with your manual intervention (rotation, and then Levels) in between or before stages.)

And now—

We Start Our Script

Begin a new Action in the Actions panel. I'll name mine "Cerebus V1 Newsprint Pt 2".

I'm going to list these steps below in a fairly perfunctory way. If you have trouble following along, please read over the first few installments again concerning original art, where I go into all of the above in a much more detailed way. The steps—

1. Hit Flatten Image. (Good in case you end up running the script on images in progress in some way.

2. Hit Control-Alt-I to bring up the Image Size dialogue. First, change the Percentage box to whatever amount of enlargement or shrinkage you want overall from the original scanned size of the image. In this case, I'm keeping it the same size, so I'll leave it at 100 percent. Then change your resolution from our scanned resolution (should be 1200) to your target resolution (should be 2400). Make sure the algorithym is set to Preserve Details, Photoshop's excellent fractal-based upscaling routine, and then hit OK.

2. Make a copy of the Background layer, and name it "noise reduction."

3. Make another copy of the Background layer, and name it "sharpened"

4. Make a blank new layer and name it "cleanup"

5. Make a new Threshold Adjustment Layer, and then turn it off.

Here's what your panel should look like so far.

Now, for the good stuff...

Next week: More newsprint scripting! Making the most of your many options.

Sean Michael Robinson is a writer, artist, and musician. See more at LivingtheLine.com.

1 comment:

Tony Dunlop said...

I don't really read these, being something of a techno-weenie, but seeing the photos of Dr. Mara remind me how much I wish she had stuck around long enough to give us her psychological take on "Reads."