From Fedora Project Wiki

Useful tips by Andrew Walton

Print design tips

A special separation trick for very dense images. ie: lots of ink on the paper.

Most printers will lie and cheat in this particular situation, be warned.

Your separation is called CMYK for a very special reason. Technically that is the order in which the printer has to put the colours down on the paper in order for the image to appear as the designer intended. So Cyan and Magenta go down first, then the Yellow.

But the yellow ink is the slowest drying of all of them, and you're putting it down on paper that's already wet with 2 layers of ink. And then you have to try and get the black down without smearing over the top of that. If your boss insists on buying cheap rubbish inks it actually never dries at all. So the printer cheats, he does the print CMKY with the Yellow last. With very dense images it's actually impossible to do it any other way, the paper is still only paper.

So if you have a particularly dense image you want mass produced talk to the printer before you do your separation. Don't talk to the sales rep, he usually has no clue. Talk to the actual offset printer tradesman that will be printing that job, let him decide if your separation needs to be CMYK or CMKY.

It's also a good idea to talk to the actual printer about what papers to use, after all, he's the one that has to try and get the ink to dry and he knows what papers will give what jobs the best results. The sales rep is just trying to quietly drain your wallet and will repeat anything that the sales rep from the paper company said even if he knows it's not true. Some papers are so alkaline that we had to stack customers jobs carefully in the backs of every one's cars and reverse them all out in to the full sun for a few hours to try and get them to dry. And I'm talking Australia here, not them cold places like Europe and the US.

CMYK colour separations

I was actually an Offset Printer for more than 20 years, computers and photography were only hobbies. Probably where my mental issues come from - heavy metal poisoning.

There's a big mistake that nearly all graphic designers make, especially if they're fresh out of TAFE with a shiny new certificate in Photoshop.

Paper is only paper, if you put too much ink on it it gets soggy and changes shape. This is really noticeable in colours such as the dark blue for Fedora and in any darker areas of the image.

The problem is that when you do a default separation your software by default will use all four colours to make black, even though one of those colours is black.

This was way before I started playing with the Gimp, maybe Gimp gets it right, I don't know. The old Quark used to do the separations correctly but Photoshop is hopeless, what you have to do is lift all the black out of your image as a separate layer first, and put it to one side as the finished black layer. Then do your cmyk separation, the black layer should be blank and the colour layers should be almost empty any where the black was.

I was a "short run speciality" printer, someone else did the 50000 posters, I did the 300 corporate invites. If you are truly after a high quality result use cmyk for the photos only and use "spot colours" for the rest of the design. It's not uncommon in my trade to get a job that's listed as "7 colours, 2 sides, + 2 varnishes and a spot varnish"

The spot varnish is a clever little trick, you give the finished print an over all gloss varnish first, then use one of the same printing plates from the job, usually the Cyan, and print a matt varnish over the top. This trick can be used to attract peoples attention to an area of the design without making it too obvious that this is what you want.

Screening your CMYK separation

Printer's inks are transparent, except for a few specialty colours. This gives the printer control over colour adjustments, the thicker he puts the ink on the darker the colour gets - within reason. This is also why the printer has to put the colours down in the exact reverse order of how your colours "should" have been lifted from the image during separation. Putting the Magenta over the top of the Cyan does not give exactly the same result as putting the Cyan over the top of the Magenta. In a four colour process print job putting the colours down in the wrong order produces an effect similar to a bad white balance shift, then the printer finds out that he can either try to make the skin tones look natural or make your company logo the right colour, not both. (refer to other comment about printers cheating CMKY)

Putting the ink on the paper in different densities all over the page however is physically impossible, in fact the ink roller systems (roller trains) on his press are designed to help prevent this from happening by accident. Instead the image gets broken up into an imaginary grid pattern of squares, where the colour is darkest in the picture each imaginary square is filled with ink but as the colour gets lighter across the picture you get smaller and smaller dots of ink in each imaginary square, making bigger white spaces in between. The graduation of colour is actually an optical illusion.

The less dots per inch you use for your image the less illusionary that graduation is. But if you make your screen too fine, too many dots per inch, you can very easily make it finer than what a simple aluminium plate can hold. The smaller the dot the less deeply the oleic emulsion gets embedded into the surface of the plate and the quicker it gets worn off again.

Printers talk in dots per linear inch, by the way, not dots per square inch. Sometimes to save confusion they'll use the term "lines per inch" instead.

The most commonly used screen size for Offset Printing (lithographic) is 133 dots per inch, this gives a fairly good reproduction and the printer can expect to get between 30000 and 50000 impressions from one plate depending on weather conditions, type of paper, and how good the operator is. This would be for most glossy brochures and fancy business cards.

For a truly high quality image you could use 175 dots per inch but the printer won't get much more than 1000 copies from the plate. Every time the printer has to line up a new plate he basically has to set up the job again and this cost will be passed on to you - the customer. So it's good for short run quality jobs but it also requires a more highly skilled printer, can only be used on small sheet sizes and it will cost more. 225 dots per inch is possible but impractical. If the quality you require is finer than that then you need to talk to a Gravure printer, they can do 2000 dots per inch but you'll probably choke when you hear the cost.

Very large sheets of paper present a new problem - paper stretch. When paper gets wet it expands. As each next colour gets put down the image underneath it gets spread a little bit bigger. Percentage-wise this expansion is tiny but across a large poster sized sheet (A2 or A1) this stretch can be up to a millimetre. A way to help hide this is to use a larger screen size, less dots per inch. And more overlap on your spot colours, the 20 microns I mentioned in the other email is only good up to A4 size. Cheaper papers usually get more stretch as well, and let's be honest, if it's 50000 posters advertising an event and they're going to be useless afterwards then it's cheap paper that you're going to be buying.

Take a close up look at the next truck or bus you see with one of those giant adverts on the side, the bloke that makes the giant stickers for them has the same problem, those screens are huge. Your newspaper also uses a larger screen because it's on cheap paper and it's a large sheet. Those special colour lift-outs you get are actually on expensive paper imported from Finland so that they can make a better quality image, rub it between your fingers and you can feel the difference, the more expensive paper has a much finer texture.

The other thing to consider with your screens is the angles of separation. Most print oriented graphics software will get this right automatically but understanding why won't hurt. If all four of your separated layers get the same screen on the same angle you will get a prominent "moire" effect across the image. Technically this shouldn't happen, until you take paper stretch into account, as each progressive colour gets put down the screen underneath it gets slightly bigger and so none of them can ever fit exactly on top of each other.

How we get around this is by rotating the screen by 15° for each layer, so the Cyan gets a screen square with the edge of the page, Magenta's is rotated to 15°, Yellow's is rotated to 30° and Black's to 45°. When I first started we used to do this in the dark room by putting a special sheet with a freznal screen over the top of the negatives.

More on dpi settings

just to save any confusion, the "dots per inch" resolution that you're working with while building and editing your image can be considered a little like the zoom level when you're working with RGB. More dots per inch is very much like more pixels per inch, it means you're more zoomed out and looking at a smaller sharper image.

This need not bear any relationship to the dpi setting when you screen your print ready files. Unless you're using default settings and letting the software figure it out itself.

It's not unreasonable to work with 600 dpi images for a sharper resolution across the page, then apply a 133 dpi screen when you're finished. It actually helps the quality if the dpi setting of your image is not a factor of the dpi setting for your screen. So if you work at 399 dpi then use a 133 dpi screen you may as well just use the default settings.

Going the other way can be a little tragic though.