How do I resize and make line art smooth when blowing up line art in Photoshop?

[arkitecht]

Hi,

I am an fine artist and want to make large giclee-style canvas prints (e.g. 30in x 35in) of my art. The problem I have is, I draw the line art and ink it on an 8.5x11in piece of paper. When I scan it at 600 dpi on an HP scanner and then try to resize it to a very large to something that's big, like 30 inches by 35 inches, for example, the line art looks "grainy" and not smooth. I use "Perfect Resize 7 Professional" which has the fractal algorithm to blow up the line art, but lines still look grainy. The plugin was very expensive too, so I hoped it would solve my problem.

So, how do I get the lines to look smooth at such a large size? Do I have to trace every hand drawn line over using the Photoshop paint tool? Is there any way I can smooth the lines out with a plugin of some sort?

Please help me!

Here are some things I tried:

- Took the scanned TIFF image and imported it into Photoshop, used the "Perfect Resize 7 Professional" fractal plugin, but the lines look grainy.

- Took the scanned TIFF and imported it into Illustrator. I converted the lines into vectors using Live Trace, hoping they would smooth out. Unfortunately they didn't.

I don't get it, guys, how do you professionals make hand-drawn art printed on very large canvases, billboards and such to look smooth?

I appreciate your advice and look forward to your help, I really need it

Thanks!

[garricks]

Hi arkitecht, welcome to GDF. I hope you'll find it useful and fun here.

We ask all new members to read the threads posted HERE and HERE. They explain how the forum runs, the rules, frequently discussed topics and our inside jokes.

PrintDriver is our grande format expert, and Bob has some great scanning ideas too. I'm sure they'll have advice for you. I'm sure they'll be along before too long.

[palo1]

You might try experimentation with Smart Objects, and vector shapes and masking. TutVid has various other tutorials that may help if the aforementioned don't.

As for creating glicee prints, why are you inking them outside of Photoshop? Hold off on the color until you scan the line art into the computer, and create masks to help with the crisp lines. It will also make it easier to create line art in Illustrator and color accordingly. Coloring in Illustrator is a little more difficult than Photoshop, but you are working expressly in vector format (no resolution loss when reducing/expanding image). Additionally, if you color it digitally, you won't have any contrast problems, and create smoother gradients and sharper lines. It takes a little more practice to master than the traditional pen/brush/whatever medium on paper method, but allows you more flexibility (plus that handy-dandy "undo" feature).

As for reducing the "grainy" feel of your line art, you can scan in your image, select the white space with your smart selection brush, lasso, and wand tools, then going to Select->Contract on your menu bar, and enter in a pixel value you feel is appropriate (dependant on the line thickness and resolution values), and hit the delete key. That should remove the "fuzziness" of the lines. You may have to go back in and repaint the finer lines, but that should remove the majority of the problem, especially if you pair it up with shapes and masks (these tools are your friends in more advanced projects).

You can also use the pen and line tools in Photoshop, and said tools are associated with vector shapes and smart objects.

Hope that helps you. Shoot me a pm if you need more insight.

Richard

[<b>]

Arkitecht, expanding on what Palo1 said, scanned images have fixed resolutions. When they're enlarged, they get blurry. Your 600-ppi scan, is just about as good as you're going to get. A scan at that resolution is going to pick up very fine detail in your line art, but still, the amount of that detail is limited, as is the sharpness of your lines and edges.

As you enlarge the scan to bigger sizes that require interpolation to even higher resolution, the software has to invent data that isn't even there. It does this through a series of algorithms that make guesses as to what those new pixels should look like based upon the surrounding pixels. The plugin you have makes its guesses based upon fractal algorithms, which can deliver marginally better guesses than the simpler built-in algorithms of Photoshop. Still, it really isn't possible for any software to intelligently fill in the missing data needed for the higher resolutions.

You're under the assumption that designers have some secret to enlarging these bitmapped scans, but your assumption is wrong — we don't.

You're also under the assumption that large-format images on billboards and large banners and posters are sharp — they're not. Billboard photos, for example, have quite low resolutions, and are very fuzzy when seen up close. It's just that they're always seen from far away and might not appear to be any bigger than a piece of paper held at arms length. Your brain knows that the billboard is big, but in your actual field of vision, the billboard is not that big, so the resolution can be low, and the billboard will appear to be sharp until it's approached and viewed at a closer distance than normal.

You didn't show us your line art, but assuming that it is truly line art — black and white with hard edges — there is actually a way to scale up the art with no decrease in resolution, but it involves a very different process from enlarging a scan. Instead, you need to create the art in a vector-based drawing program, like Adobe Illustrator.

This would normally mean, drawing the line art on paper, as you've done, then scanning it in, as you've also done. The difference would be that you would import the scan into Illustrator, then redraw it (essentially trace it) with vector-based lines, shapes and fills. Vector art is resolution independent since it does not depend on pixels and depends, instead, on mathematical equations that remain constant regardless of the size. For example, a scanned images of the letter "A" when blown up to the size of a billboard will have very fuzzy and irregular edges. A vector-based letter "A" (nearly all digital typography is vector-based) will still be sharp.

[sully1251]

There may be a way to do what you are trying to do—or at least coming close.
I'm going to suggest combining a pre-computer method with a computer production method. Some of the other old wizards may have some additional thoughts—

Without seeing your work, I'm having to assume a few things:
I am assuming that you are producing fine-art quality line drawings that are solid black and white— and that you like the textures and the effects of conventional drawing tools, such as brush work and ink. I'm guessing that you are working 8.5 x 11 for convenience in scanning or maybe drawing.

If this is the case—can you try working about 11 x 14 or a little larger? If so, enlarge the drawing to about two thirds of the eventual size with a PMT, to about the 18x 24 size range (or 16 x 20 range). A PMT is a photo mechanical transfer. That amounts to a good quality photostat. This PMT should be able to maintain the look of your art at a 150 % enlargement. So, at this stage you have your black and white art enlarged to interim size. This interim size could be scanned at a 600 dpi (to maintain clean edges) and printed as a giclee print at about the 20 x 30 to 30 x 38 size range you are looking for. Typically, a giclee 4 color print needs 200 dpi plus—preferably about 250 dpi.

All you production experts around here will probably find a few holes in this but we did stuff similar to this before using computers. The reasoning for the interim size is to get a much larger, acceptable line drawing without dealing with pixels.

And the trick is finding a place that can shoot a PMT. I'm sure some of us old guys know of a couple of possibilities.

The sizes and percentages are all ballpark numbers.

[palo1]

Originally Posted by sully1251

I'm guessing that you are working 8.5 x 11 for convenience in scanning or maybe drawing.

I'm also assuming that it is more necessity, as he probably doesn't have anything larger than a 8.5"x11" scanner.

As <b> said, you can bring into Illustrator (which you mentioned that you have done) and still use Live Paint/Live Trace. As per the problem you mentioned about loss of quality of sharp lines as Illustrator tries to compensate for the natural ink bleed and the lossy (the naturally-occuring algorithms that try to fill in that missing data during the image capture once saves into a file format) by adding extra vector points.

There are ways around this: I am assuming that, when you import the image to Illustrator, you are working on it after you resize it to larger than 8.5"x11"; Illustrator tries to create as simple paths as best a possible by using as much information it can get from the image--the more information in an image, the more complicated the paths, the more ragged the lines.

As such, I would actually try reducing the dimensions of the image so you have less erroneous detail, as well as fine-tuning the amount of detail that is interpreted by the livetrace/paint presets. You will probably have to go in afterwards with the pen tool to refine the points and eliminate unnecessary points that are causing ragged lines, but once this is accomplished, you can enlarge or reduce the image without loss of quality (as the image will be vector).

As for large canvases and billboards, what <b> said again: Large canvases such as the dimensions you are listing are usually around 300 dpi, but I've seen this number go higher (but rarely lower, and never below 200 dpi). 4'x6' typically have a resolution of 150-200 dpi, and billboard dpi can range anywhere from 12 to 600 dpi (depending on the size). A good rule of thumb is for every 1" to 1' conversion, the dpi usually remains constant. That being said, 1" at 300 dpi for traditional print is 1' at 300 dpi for a traditional billboard size (12'x24'), or 25 dpi (ballparking-- actual dpi varies with printers/optimal viewing range/size of billboard).

That, or course, is assuming that you were working with raster images, rather than vector. I already made this point in my previous post.

[palo1]

Additionally, this ratio is not for every project.

[<b>]

Sully, I haven't heard mention of a PMT for at least 25 years. Does Kodak or Agfa even still make stat paper?

I used to use the process you've described to make larger display-sized headlines from Letraset rub-down type. I'd rub down the largest size type I had, toss it in the stat camera, then blow it up to the size I needed. From there, I'd use a Rapidiograph pen and an X-Acto knife to clean up the edges and sharpen the corners. Sometimes I'd even make a PMT from the PMT for the finished mechanical, but geech, that was back before I even had whiskers, let alone grey hair.

I don't know where you'd find a stat camera these days. Maybe one of the prepress or large format people here would know. I suppose it would be possible to do something similar with a larger-format xerographic copying machine and a harder-surface paper.

[palo1]

Originally Posted by <b>

I don't know where you'd find a stat camera these days. Maybe one of the prepress or large format people here would know. I suppose it would be possible to do something similar with a larger-format xerographic copying machine and a harder-surface paper.

If they were going to go to this extent, wouldn't be better advised that they would take their work to a large-format scanning company that uses something like a Cruse scanner that has color balancing and shadow exclusion?

Just my opinion.

[Bob]

I think I'd like to see a small sample of the line work in question. If it is JUST the linework and you want to blow it up roughly 400%, the best way is to scan it in greyscale first. THEN upsample that scan to 400%. Now stick a Threshold adjustment layer on top of that and find a good 'typical' detail area to view at 100% size. Okay? Now back to the scan layer, time to WHACK IT with Unsharp Masking, maybe a couple of hits. Keep an eye on the preview.

This is my standard way of handling fineline artwork like etchings, crosshatch work, etc. when going from a 300 ppi scan to say, 1200 dpi black and white (1 bit) line. It works better than the theory says it should. Go figure.

If you're going to Giclee, you probably want to stay in greyscale though -- I don't know if you're planning to layer this with a wash or colour layer. In any case, you probably don't want to flatten it down to a 1 -bit line image unless you're compositing in a layout app (like ID). In which case, maybe lose the Threshold layer and add a Curve adjustment for 'anti-aliased' looking edges. Tweak the curve to suit.

Sorta.


This works great for freehand type of lineart