Pre-press question re: "nickelling"

I’m having problems with some curves “nickelling” in EPS files when the files are being ripped for final output (to film or DTP). As the jargon implies, instead of say a smooth round circle, you get something that looks like a nickel – a bunch of short, straight lines forming a multi-sided shape, not a true circle.

Strangley, it’s not happening consisntently, only with some files. The files themselves are not corrupted, and most of our printers (we outsource) have no troubles with them.

The files were created and save in Adoe Illustrator as EPS files.

This only happens to a few files after they have been ripped. Printing them directly from Illustrator - fine. Creating a PDF - fine. Converting to web graphic - fine.

I’m thinking it’s something that’s happening in the postscript stages. Some weird preference maybe that is not exporting the curves properly. I suspect there is a document setting or some kind of object attribute in the files that are goofing them up, but I’ll be darned if I can find it.

Any ideas what I should check?

I don’t have any really good ideas, but here’s one thing you could check: Does it happen with all kind of curved lines, or just some kinds? PostScript has different kinds of curves: circular arcs (the arc* commands), which may be distorted into ellipses by the transformation matrix; and Bezier splines (curveto and rcurveto). Maybe the conversion is handling one of these well and the other poorly.

Do both of these curves look smooth? (The one with a dot in the middle is an actual circle, generated with arc; the other one is made of four joined splines.)


%!PS-Adobe-3.0 EPSF-4.0
%%Title: arc/curve test
%%BoundingBox: 0 0 200 100
%%EndComments
%%EndProlog

0 setgray
/r 40 def /da 22 def /db 0 def

newpath 50 50 r 0 360 arc stroke newpath 50 50 2 0 360 arc fill

newpath 150 10 moveto
  da db r db sub r da sub r r rcurveto
  db neg da da r sub r db sub r neg r rcurveto
  da neg db neg db r sub da r sub r neg r neg rcurveto
  db da neg r da sub db r sub r r neg rcurveto stroke

%%PageTrailer
showpage

%%Trailer
%%EOF

<checks pocket change>
Looks pretty round to me!

Is nickelling some arcane printers term having to due with uneven buildup of plating around certain types of curves?

What version of Illustrator?

Used to be in older versions of Illustrator, you had to tell it the document resolution. Set it too low and output the file on an imagesetter the curves would no longer appear smooth.

I think this has to do with your flatness setting. Illustrator has a control that will allow you to approximate a curve, basically reducing the complexity of the curve and requiring less memory from the printer (to avoid a limitcheck error). But sometimes the result can be undesirable, making the curves appear to be made up of edges. The higher this setting… the crappier it looks.

I can’t remember the last time i’ve had to deal with this issue… but i swiped this from Illustrator’s online help (search: flatness):


Approximating curved paths

Curves in artwork are defined by the PostScript interpreter as small straight line segments; the smaller the line segments, the more accurate the curve. However, as the number of line segments used increases, so does the complexity of the curve. Depending on your printer and the amount of memory it has, a curve may be too complex for a PostScript interpreter to rasterize. In this case, a PostScript “limitcheck” error can result, and the curve won’t print.

The Flatness option gives you control over how closely you require the printer to approximate curves. You can set the following Graphics option in the Print dialog box:

Flatness

Specifies how closely to approximate curves in the artwork. A lower setting (toward Quality) creates more, smaller straight line segments, more closely approximating the curve. A higher setting (toward Speed) results in longer and fewer line segments, creating a less accurate curve, but improving performance.


Hope this helps!

Yeah, there’s a PostScript setting called “flatness” that’s responsible for the effect you’re seeing. Essentially, curved lines are always represented by a raster image processor (RIP) as a sequence of very short straight lines (think about it – pixels are square, so it’s never really possible to draw a curve). The higher the flatness setting, the longer these short straight line segments are. When a curve described by the PostScript instructions would require the calculation and drawing of more points than the raster image processor (RIP) has memory for, a PostScript “limitcheck” error with “lineto” or “fill” or some other drawing operator as the offending command will occur. Increasing the “flatness” setting for the device will cause the curve to be subdivided into longer straight lines, each of which is (with any luck) within the capabilities of the device. So higher is better, but only up to a point, as higher values can also can lead to the effect you’re seeing. If the application you’re using to create the graphic in question offers an option to set the flatness in the print dialog box (Illustrator does), experiment with a lower setting. The ideal setting is the highest one that allows the image to print on the target device without the effect you’ve decribed (higher settings allow the image to print more quickly).

The reasons some printers have this problem and others don’t are that (a) this is likely to be a problem primarily on imagesetters and other very high resolution devices (>1200 dpi), especially if they’re older devices with limited amounts of RAM, and (b) the default flatness setting for the RIP is a persistent parameter that output service providers often tweak to provide the best results for their devices with their customer base, so even if two providers are using the same device, they may have the default flatness setting on the RIP at different levels.

If you’re unable to find a setting that allows the document to print successfully on a given device without the effect you describe, the other strategy here is to use the option in Illustrator to Split Long Paths. This basically breaks up the curves as drawn in Illustrator into multiple shorter curves, making it easier for the RIP to digest each one. The downside is that it’s much more difficult to edit an Illustration after using Split Long Paths, since there are now tons of short, arbitrary curves where you used to have nice long flowing lines.

There’s a decent explanation of all this here, with some examples.

DING DING The winner! That was the problem. Sorta.

My document output resolution was 2400 (hence my confusion – why the hell was it nickelling???) I have a newer version of Illustrator on one computer and older version on the other. The problem was that I DID have to correct output resolution for the document… but older versions of Illustrator also have an output tag for each object (which seems pretty silly to me) under its attributes.

I selected the “bad curve” had to go to Window > Show Attributes and sure enough the output settings for the naughty curve was a mere 100.

Bah! I have no idea why the output was so low. You’d think giving the document a high output resolution would override over the stupid setting on a single object, but nooooooo!

Thanks everyone. I looked into everyone’s suggestion.

I love this place!

I was interested in how this term came about too. Looking at a couple of graphic artists sites on google, one said that Canadians would understand the meaning of the term.

Apparently that’s true, Here’s a ridiculous example of an older Canadian nickel. The current nickel (with a beaver on it) doesn’t have facets anymore.

_
_

You’ve gotta be Canadian, eh?

It’s Canadian pre-press jargon? :smack:

Wait, there has never been an American coin that has been faceted?

Okay, so what is the pre-press term in the U.S. for “nickelling”?

Our one dollar coins (affectionately called Loonies) are still faceted, pennies were in the 90s too (made the easier to distinguish by touch when you’re reaching in your pocket).