Wednesday, 22 February 2017

Paper to Pixel to Paper Again: Part 7

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 7
Scripting & Upscaling


This is the seventh 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.


Last week, we left off in the middle of making an "Action" command in Photoshop, that we will use to adjust and prepare our original art scans.

Here's what mine looks like, after having dumped the less desirable color channels and converted to grayscale—

(You can see there are a few extra moves in there I could delete if I wanted to really clean it up, but  none of them will slow down the script)

Now we're going to make one last move before we upscale our art to the actual resolution we'll be working at. I'm going to make a Levels adjustment to knock out a bit of the paper color and see if we can enrich the black a bit prior to the upscale.

If you still have your original art with you, scan your eyes over the page and try to find the finest detail that you can pick out with your eye, whether that be a dinky little line or spatter or a spatter tone. Now zoom in on that area of your scan. Really, really far, like 400 to 700 percent, depending on the resolution and size of your monitor.

Now hit CTRL-ALT-L to bring up the Levels command.

We're going to talk about this window a bit, as we'll be spending an awful lot of time here. (Adobe's official explanation of the Levels histogram, which is a very concise unpacking, is here.)

The above histogram represents the distribution of tonal information within the layer that you're currently modifying. See those two peaks? Those represent the "black" of the ink (on the far left) and the "white" of the paper (far right). And that valley between them represents all of the gray information distributed throughout the image. The lowness of that valley indicates there's not much information there at all, compared to the spikes for the "black" and the "white" of the paper. Which is what we'd expect to see for a grayscale line art scan.

What we're going to do now is move the black point of the image (circled in magenta above) all the way over to the edge of the first peak, effectively making our "black" ink actually black. And then we'll move the white point of the image (circled in lime green above) to the right to meet the edge of the paper peak, essentially knocking out some of the paper color of the image.

While I'm making these adjustments, I'm watching two things carefully. In the areas of black detail (i.e. dense hatching), I'm looking to make sure moving the black point isn't causing any of those details to fill in at all. I want the black point adjustment to only impact the solid black areas and not cause any other effect. Similarly, as I adjust the white point, I'm watching to make sure that the very finest details aren't shrinking at all. All we're trying to do is prepare the page, not get rid of any information. (Yet another reason to be conservative here—we're going to be running this script on every page. So hopefully this is a representative sample of the whole!)

Okay, after you've blackened the blacks and knocked out a bit of the paper color, grab the Mids control (circled in cyan above!) and move it about a bit. This controls the overall exposure (or gamma) of the image. Moving it to the left (towards the black point) will lighten up the page overall. Moving it to the right (towards the white point) will darken up the page overall. Play with this and see if any additional details becomes visible as you move it to the left, or if the blacks become richer without losing any detail as you move it to the right. (Later, when we're dealing with photo negatives, this will be an invaluable technique.) After you've settled on a good home for your Mids you might be able to back off a bit on your other adjustments.

Here's where I ended up on mine.



Now, turn off your Actions recording and let's make sure your adjustment is helping your image. 

ALT-CTRL-Z steps backwards/undoes that last thing you did in Photoshop. SHIFT-CTRL-Z re-does that last thing. Skip around in your image, zoomed in, undoing and redoing the last step, and taking a look at the effect. And get used this these shortcuts. We'll be using them a lot. 

Now that we're prepared our scan, we're going to upscale our image to our final resolution and size. Hit ALT-CTRL-I to bring up the IMAGE SIZE menu. (Or go to Image-> Image Size. But really, the keyboard shortcuts save you valuable time. Best to take a bit to learn them, especially for things you'll be doing over and over again.)

Here's the Image Size menu. All of the darker gray boxes on the right are breakout options boxes. If you click on the arrow on the right of each, they'll open up additional options.

First, we're going to change our Percent number. Our overall artwork needs to be shrunk to 62.4 percent of its original size, so that's the value I'll enter here. (In the original run of the Cerebus book, the artwork was usually reduced to 60 percent of its original size, but in the remastered books it's slightly larger than that).

Next we'll change the Resolution to our target delivery resolution of 2400 Pixels per Inch. You'll notice that when you change this value, your Percent value changes as well, in this case, multiplying your percent reduction (62.4) times the resolution change (4), leaving us with 249.6 Percent.

Lastly, we're going to change the type of resampling that Photoshop applies to our image. Select Preserve Details from the menu. And then change the "Reduce Noice" slider to zero. (You only need this if you're working with a color image, especially one suffering from JPEG compression or some other kind of artifact that might otherwise get caught up in the upscaling algorithm).

(This is already an overly long entry, so I'm loathe to get too far into the differences between these resampling methods. But I think it's important to note that Preserve Details is fundamentally different from Bicubic Smoother, which is the other method you might use for enlargement. Preserve Details is a fractal method of interpolation that is ALMOST ALWAYS the way to go for upscaling. Exceptions would be upscaling certain kinds of line art with mechanical tone where the mechanical tone has not been adequately captured by the initial scan, in which case Bicubic Smoother is the better way to go)

Here's a screenshot of my final settings.

And now we click OK and wait for a minute... 

(Still waiting... now you see why I told you to make all of this an Action! So you don't have to stare at the clock or check the news or check your pointless Facebook feed and find out no one loves you and everyone is clearly happier than you are and hey look there's a video of a man catching a trout with his face, better check out the comments, and it looks like Sylvia's broken up with David again and...)

O hey, it's done!

A Brief Detour—Why Do We Upscale?

As I mentioned back the first and second installment of this series, just because we plan on delivering 2400 ppi 1-bit bitmap images to the printer, doesn't mean we have to scan at that resolution. Grayscale and color scans have much more data per pixel than a 1-bit image, and we can use this data to upscale to our desired resolution, with very little downside at all.

Below you'll find a comparison image I've made with a small portion of the page we're working with. The first image has not been upscaled, only converted to a 1-bit bitmap. The second one has been upscaled exactly as detailed below, and then converted to a 1-bit image, with nothing else done to the image.

As you can see, the upscaling has achieved the goal of bringing us into a much finer resolution space with much smoother edges and details, without affecting the balance of details and dense areas in any way. If you want to make a comparison image yourself, then make a version of your page upscaling with "Nearest Neighbor," then copy and paste the result onto a new layer of your working image. 

Back in The Present

Now I'm going to go to the Layers panel and right-click on the layer with our image, and click on "Duplicate Layer." Name this new layer "Sharpened." We're going to make our final adjustments on this copy, so as to maintain the original below.

Next, with our Action still running, click the "Make New Layer" button at the bottom of the Layers panel. Name this new layer "Cleanup." Next, go to Layer -> New Adjustment Layer -> Threshold. This will create a way to preview what our final 1-bit image will look like after we convert it. But for now, turn off this adjustment layer by clicking the Eye icon next to the layer name.

Lastly, click on our "Sharpened" layer and then turn off your Action. We're going to spend a bit of time now playing with different possibilities.

And here's our script and layers panel so far!

So, uh, cliffhanger?? Will we sharpen? Will all of our detail remain? Will the printer forget to clean the press and drive our book to Dotgainville? Find out NEXT WEEK!

(To download the top portion of the page I'm working with here, click this link and then click on "full res version," then the download button. Then feel free to follow along!)

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


Jeff Seiler said...

At the risk of being redundant,


I wish I could understand even a quarter of this wizardry.

You know, a coupla hundred years ago, my kind (dumb Luddites) probably would have burned you as a witch.

Just sayin' astound me!

Dave Kopperman said...

I cannot tell you (Sean) how goddamned USEFUL this all is - and your timing is perfect, since I'm in the middle of a large pen & ink project that I was fretting about reproducing properly. Normally, I'm not such a stickler, but since this one's going on the cover of a vinyl LP, I really want the final to get all the detail and clarity you're delivering here.

Sean R said...

Jeff--don't hesitate on my behalf :)

Hey Dave! Glad to hear it's been useful. Check out next's week's sharpening bit (tried to get to it this week but got caught up in the details--like a building on the horizon that you don't ever seem to reach...). The only other thing I can tell you is, the 2400 ppi delivery thing really needs to be confirmed by the prepress staff of the printer or pressing plant. They most likely have their RIP engine set to automatically downsample to 1200 ppi, so they'll need to know to turn that off. (Though if you're not using any mechanical tone, a 1200 ppi downsample might be fine.) Just put a note in there letting them know not to do so, and make them reply in agreement :) Nicely, but in ALL CAPS!

Unknown said...

Sean R and Dave K - YES! It's like Sandeep's potato gag: "Yeah yeah yeah and then there's no potatoes I really hate that." There is so little fine-line pen and ink stuff being done today that even the professional pressmen don't really know how deep into the page you have to go to get everything. As much as possible, make sure to get printed proofs and if it's 1200 and not 2400 make a stink about it.

Fine-line inking is only going to get more rare as almost everyone heads over in the direction of Computer 101.

Dave Kopperman said...

I'm definitely looking forward to sharing it with the room, as it were, when I'm done.