Resolution

[Gudern]

Xpian you are right - since the very beginning I have been talking about PPI and not DPI.
I even explicitely said in the first post that I assumed no impression to avoid this confusion.

Redrobes and Waldronate thanks, you are on topic because what you wrote seems to have answered my original question.
I'll sum it up and please correct me if I misunderstood something.

- There is a physical support for the map (a screen) which is approximately a square 30 cm by 30 cm - it may be bigger or smaller but I take this example which corresponds to a diagonal of 42 cm.
- People generally watch it at a distance of around 40 cm. From Redrobes tutorial and Waldronate's remark follows that the eye resolves then approximately 200 PPI (or 80 ppcm)
- So the size of the map in pixels directly and uniquely follows from the 2 numbers above. It is 2400 x 2 400 pixels. If it is less, the eye will see individual pixels and the map will be pixelated. If it is more, it is useless because the eye won't resolve the additional pixels so that a bigger size is an overkill (and useless work)
- If I want to represent a continent 3 000 km x 3 000 km, each pixel represents 1,25 km in reality
- Follows that I have to draw the map so that everything that is bigger than 1,25 km must be shown because its size on the map will be AT LEAST 1 pixel.

Of course all that assumes that I do a map where the scales are real and not a map where there is no relationship between the real size of represented objects and their representation on the map. And this is indeed my idea.
So is it right that the answer on my question is : "In your case take 2 400 x 2 400 pixels. If your screen is bigger or smaller than 30x30 cm, just apply a proportionality" ?

[Redrobes]

I would agree that 2400 x 2400 sounds about right as the final res for your project.

If you continue with the tutorial post to antialiasing, I recommend people use twice the figure they need (4800 square) and do the working art at that resolution and then for finals, make a copy, just slightly blur it so that you get a two to one pixel blend, then resample it down to half (i.e. 2400 square) for the final result. But keep the working copy and make new changes on the higher res version. Its not strictly necessary, but if you have multiple close lines in your map then you might hit the aliasing issues unless you do this. If you specifically intend to use hatching or similar shading techniques then I would definitely recommend it.

[Carnifex]

NOTE: there's a whole discussion to be had about different types of process color screens, specifically the difference between the standard 4-color process you see in magazines, newspapers, packaging, most posters, and most books, versus the stochastic screen used in inkjet or dye-sub printers. And the standard 4-color process is measured, not in DPI, but in LPI or lines per inch. But the actual guidelines are basically the same: you can use graphics that are about half the LPI of the printing press in PPI...and maybe up to the same LPI to PPI ratio for really fine, sharp lines.

You actually need the double dpi compared to the lpi. So a printer raster of 133 lpi demands a 266 dpi image. As a simplification you usually recommend 300 dpi which is normally enough. (Note that line-art images (pure black and white) are not rasterized at all (recommended dpi of 1200 for best quality).)

[Redrobes]

You actually need the double dpi compared to the lpi. So a printer raster of 133 lpi demands a 266 dpi image. As a simplification you usually recommend 300 dpi which is normally enough.

This bit is true. An inkjet, dye sub, laser or dot matrix printer will need to print a coloured dot followed by a space followed by a dot again to produce lines so two dots per one line.

(Note that line-art images (pure black and white) are not rasterized at all (recommended dpi of 1200 for best quality).)

This is not so true. Those same kinds of printers will rasterize the lines as part of the printer driver. Only pen plotters and other very specialized hardware will not have to rasterize a vector image to draw it. As such for almost everyones case on this forum, every image will eventually become rasterized.

The advantage to vector art is that it is rasterized at the final device stage and not earlier. But if you run with a resolution that is higher than the expected printing stage then there is not much reason to say that one is better than the other. If you dont know what size image and what device your going to use to print it, or show it, then there is advantage to vector art by not fixing its resolution too early. But then there are other advantages to rasterized images too.

[xpian]

@Carnifex - you're right about the halftone screen numbers. I was mis-remembering. The guide I was thinking of was 1.5 times in pixel density versus line density for greyscale images--using 225 ppi when printing on a 150 lpi press, for instance. Any more ppi is wasted information that the printer cannot possibly produce. For color halftones, the advice is as much as double the ppi resolution as compared to the lpi. The stochastic screens of inkjet printers are another matter. I found an online article that does a great job of showing why. http://the-print-guide.blogspot.com/...pi-vs.html?m=1

@redrobes - you're right about everything being rasterized. Virtually every printer people encounter has its own internal machine raster and will not only rasterize vector content, but re-rasterize pixel content. Incidentally, back in architecture class I actually occasionally worked with those true vector printers, plotters that actually used pens you'd load in them and could draw truly smooth lines. When I worked at a sign and banner shop, we had a vinyl cutter that worked the same way, where you'd feed a vector file to the machine and the little blade would smoothly slice lines across the roll of adhesive PVC.

[Carnifex]

This bit is true. An inkjet, dye sub, laser or dot matrix printer will need to print a coloured dot followed by a space followed by a dot again to produce lines so two dots per one line.

This is not so true. Those same kinds of printers will rasterize the lines as part of the printer driver. Only pen plotters and other very specialized hardware will not have to rasterize a vector image to draw it. As such for almost everyones case on this forum, every image will eventually become rasterized.

The advantage to vector art is that it is rasterized at the final device stage and not earlier. But if you run with a resolution that is higher than the expected printing stage then there is not much reason to say that one is better than the other. If you dont know what size image and what device your going to use to print it, or show it, then there is advantage to vector art by not fixing its resolution too early. But then there are other advantages to rasterized images too.

Not sure we understand eachother correctly?

I have been working professionally with printed matters for the last 25 years. Specializied in final art and teached graphic design for the last 7 years. Worked my first 5 years in a print shop ripping stuff.

A line art tiff image is not rasterized but printed in full resolution in an offset printer (an offset ctp (computer-to-plate) printer uses maybe 2400 dpi or more - a laser printer maybe 300 or 600 dpi). 600 dpi for an line art image is usually enough to make it look good. There are some printers that might rasterize everything but definitively not an offset ctp or quality laser.

Comment: Maybe you meant that vector art is not printed exactly vectorized - maybe you can call it converted to the printers dpi? But converted to a raster such as a color half tone? No.

A good example of a thing not rasterized is text and vector graphics. If you use a magnifying glass you easily see the difference between a rasterized image and (for example) text.

[Carnifex]

@Carnifex - you're right about the halftone screen numbers. I was mis-remembering. The guide I was thinking of was 1.5 times in pixel density versus line density for greyscale images--using 225 ppi when printing on a 150 lpi press, for instance. Any more ppi is wasted information that the printer cannot possibly produce. For color halftones, the advice is as much as double the ppi resolution as compared to the lpi. The stochastic screens of inkjet printers are another matter. I found an online article that does a great job of showing why. http://the-print-guide.blogspot.com/...pi-vs.html?m=1

.

Not exactly true. 1.5 times the lpi might be ok but 2 times are recommended.
Here's from your linked page:
"Here the image ppi/dpi is twice the halftone screen resolution (lpi). As a result the halftone no longer resolves the individual pixels of the original - just the tones they represent.This*ideal*required original resolution can be represented by the formula:*2 X lpi = ppi @ 100% reproduction."
This is a greyscale example and the same is true for color.

[xpian]

@Carnifex - The author of that web page has several levels of recommendation for PPI to LPI ratios. There is a section for greyscale and a section for color, and his recommendations are slightly different. He has *minimum*, *ideal*, and *maximum* categories. Having looked over the page in question, it seems to be in line with both my experience as a printer and what I've read from many other sources. I'll just refer people to the link, rather than repeating here what that (excellent) source says.

The above discussion is regarding traditional "4-color Process" halftoning, such as is found in newspapers and most magazines. One of my points, which we have not really been discussing, is that many people printing things these days are doing it using desktop inkjet printers and that these printers use an entirely different form of halftoning. In stochastic screening as with inkjets, the dots are much smaller and sprayed in a semi-random pattern where density of tone is achieved by clustering dots tightly rather than varying the size of the dots. I'm sure many people already know this, but the upshot of this difference is that the PPI resolution recommendations are different. If I look at the inkjet printer on my desktop, it says it has a "1200 DPI" resolution. I wouldn't want people to think they need to send files to it that are 1200, 1800, or 2400 pixels per inch. That would be silly. When printing with stochastic screens, especially when concerning smooth colors as in photographs or paintings, the PPI of your file can be much lower than the DPI of your printer.

When it comes to your discussion of printing line art TIFFs on a 2400 DPI CTP printer...I think that the argument is highly technical and that both sides are, perhaps, correct in their own way. I can't really speak for Redrobes, but I think we're saying something very similar: every output device has its own, built-in internal raster. The CTP printer you mention has a 2400 DPI raster. Every file sent to it, whether that original file is raster or vector, gets converted to the machine's internal raster at the time of output. How could it be any other way? Since we're talking about line art, we're only discussing black and white pixels. As long as the file's resolution is less than or equal to the printer's DPI, in this case, there should be no loss of quality. There will be no "re-sampling" evident in the image when it's printed out. It will appear to be exactly the same, pixel-for-pixel, as the file that was sent. I think maybe this is what you mean by "A line art tiff image is not rasterized but printed in full resolution in an offset printer"--there's no evidence of upsampling, downsampling, or any other image degradation so long as the output device has super-high resolution.

"A good example of a thing not rasterized is text and vector graphics." -- I think there's some confusion here with what is being meant by both sides. Redrobes (I assume) and I are not saying that the text and vectors are being rasterized at the computer before being printed. Not in the same way you might pick a text layer in PS and choose the "rasterize type" command, no. (This can happen, of course, if the designer has a need to do things to the text in PS) Most of the time with fonts and vectors, whether they be in an Illustrator file, a PDF, a Word doc or whatever--most of the time, this info is sent to the printer as math. Often in the PostScript language. But, before the printer can print it out, the printer itself *must* rasterize the vector data to its own internal, machine raster defined by its hardware. And this is often so finely detailed today that you'd never be able to see the individual dots or pixels, even with a magnifying glass (if we're talking about line art, still). The only way this is not true is when talking about true vector printers, such as architectural plotters that use pens and roll the paper back and forth.

[Carnifex]

For this reason, professionals in the printing industry know that your raster graphics don't need to be the same resolution in PPI as the printer has DPI. Especially if we're talking about full-color graphics. A good rule of thumb is that you can get away with your PPI being about half your DPI, at the same size. For instance, if you know that your printer does 600 dpi on a US LTR size page, you can easily set your Photoshop file to be 300 ppi at 8.5 x 11. You'll know that each of those little PS pixels is going to have to be represented by several of the printer's dots, so it makes sense for the DPI number to always be higher than the PPI of the image. Making your image any higher resolution than that is almost certainly just wasting file size, and because computers and printers slow down when handling big files, wasting time. I've personally done many, many tests of this on lots of different kinds of printers over the years. You'd be surprised how low you can drop the resolution of your digital file and still have the print look *identical* to a higher-resolution digital file.

The main exception to this is line art, or pure black-and-white graphics with sharp edges and hard lines. Generally speaking, you might as well make those the same resolution in PPI as your printer has in DPI.

NOTE: there's a whole discussion to be had about different types of process color screens, specifically the difference between the standard 4-color process you see in magazines, newspapers, packaging, most posters, and most books, versus the stochastic screen used in inkjet or dye-sub printers. And the standard 4-color process is measured, not in DPI, but in LPI or lines per inch. But the actual guidelines are basically the same: you can use graphics that are about half the LPI of the printing press in PPI...and maybe up to the same LPI to PPI ratio for really fine, sharp lines.

I am sorry but you got this part is (mostly) wrong. Just stick to 300 ppi in your final print size. Please don't tell people anything else.

Note: yes, there are exceptions to the "300 ppi/dpi" rule.

[Carnifex]

@Carnifex - The author of that web page has several levels of recommendation for PPI to LPI ratios. There is a section for greyscale and a section for color, and his recommendations are slightly different. He has *minimum*, *ideal*, and *maximum* categories. Having looked over the page in question, it seems to be in line with both my experience as a printer and what I've read from many other sources. I'll just refer people to the link, rather than repeating here what that (excellent) source says.

The above discussion is regarding traditional "4-color Process" halftoning, such as is found in newspapers and most magazines. One of my points, which we have not really been discussing, is that many people printing things these days are doing it using desktop inkjet printers and that these printers use an entirely different form of halftoning. In stochastic screening as with inkjets, the dots are much smaller and sprayed in a semi-random pattern where density of tone is achieved by clustering dots tightly rather than varying the size of the dots. I'm sure many people already know this, but the upshot of this difference is that the PPI resolution recommendations are different. If I look at the inkjet printer on my desktop, it says it has a "1200 DPI" resolution. I wouldn't want people to think they need to send files to it that are 1200, 1800, or 2400 pixels per inch. That would be silly. When printing with stochastic screens, especially when concerning smooth colors as in photographs or paintings, the PPI of your file can be much lower than the DPI of your printer.

When it comes to your discussion of printing line art TIFFs on a 2400 DPI CTP printer...I think that the argument is highly technical and that both sides are, perhaps, correct in their own way. I can't really speak for Redrobes, but I think we're saying something very similar: every output device has its own, built-in internal raster. The CTP printer you mention has a 2400 DPI raster. Every file sent to it, whether that original file is raster or vector, gets converted to the machine's internal raster at the time of output. How could it be any other way? Since we're talking about line art, we're only discussing black and white pixels. As long as the file's resolution is less than or equal to the printer's DPI, in this case, there should be no loss of quality. There will be no "re-sampling" evident in the image when it's printed out. It will appear to be exactly the same, pixel-for-pixel, as the file that was sent. I think maybe this is what you mean by "A line art tiff image is not rasterized but printed in full resolution in an offset printer"--there's no evidence of upsampling, downsampling, or any other image degradation so long as the output device has super-high resolution.

"A good example of a thing not rasterized is text and vector graphics." -- I think there's some confusion here with what is being meant by both sides. Redrobes (I assume) and I are not saying that the text and vectors are being rasterized at the computer before being printed. Not in the same way you might pick a text layer in PS and choose the "rasterize type" command, no. (This can happen, of course, if the designer has a need to do things to the text in PS) Most of the time with fonts and vectors, whether they be in an Illustrator file, a PDF, a Word doc or whatever--most of the time, this info is sent to the printer as math. Often in the PostScript language. But, before the printer can print it out, the printer itself *must* rasterize the vector data to its own internal, machine raster defined by its hardware. And this is often so finely detailed today that you'd never be able to see the individual dots or pixels, even with a magnifying glass (if we're talking about line art, still). The only way this is not true is when talking about true vector printers, such as architectural plotters that use pens and roll the paper back and forth.

Can we agree to speak about normal offset printing?

While I am guessing you work or have worked with this but you obviously got some things wrong. I don't claim to know everything but you should really check these things out more carefully. For example: Telling people that it's ok to print stuff in less than 300 dpi/ppi is just wrong.

And you should also check out what postscript is and how it works. Before you tell other how you think it works.