Dan Margulis Applied Color Theory

Dots v. Spots

Date: Fri, 11 Apr 2003 11:32:54 -0500
   From: "Howard Smith"
Subject: Offset lithograph dots

    When a commercial printer refers to "dots", is he (or she) referring to halftone dots or to the artifacts produced by screening?  It seems that the number of artifacts per linear inch is the same as the lpi, while the number of halftone dots per linear inch appears to be--what?--eight times the lpi?
 
   It's a simple question, but one that does not appear to have a published answer.  Ask ten people and you get ten different answers.  Which reminds me of a second question.  What are the artifacts called if they aren't called dots?  No one ever seems to call them "artifacts".
   
Howard Smith
________________________________________________________________________

Date: Fri, 11 Apr 2003 16:56:19 -0000
   From: "Howard"
Subject: Inkjet printer dots

    An imagesetter uses up to 256 spots to produce a single halftone dot, and the halftone dot in turn represents 1 byte of image data.

But what about inkjet printers?  With the stochastic printing used by inkjets all the dots are the same size which means that color intensity must be represented by the number of dots printed per byte.  Is this the case?  If so, does an inkjet printer use up to 256 dots of ink per byte, or does it just approximate the percentage of dots required to match the percentage of color represented by each byte?  Finally, how does an inkjet printer print a dot,of magenta, for example, when it uses both magenta and light magenta ink?  Does it mix and match or does it produce two separate clusters of dots to represent one byte of magenta?

    It's the simple questions that will do you in.

Howard Smith
________________________________________________________________________

Date: Fri, 11 Apr 2003 16:28:35 -0400
   From: Loring Palleske
Subject: Re: Inkjet printer dots

Not all inkjets produce the same size dots. All the piezo based systems  have variable dot technology.

If you have a 300 dpi (ppi actually) image and a 2400 dpi printer, you  should have ((2400x2400)/(300x300)) 64 dots representing each pixel.  How much memory does a pixel use? greyscale is 8 bit (is there 4 bits  to a byte? - can't remember) rgb is 24 bit - inkjets generally deal  with rgb.

a dot of magenta is printed with magenta. a dot of light magenta is  printed with light magenta anything in between gets dithered.

Regards,

Loring Palleske
Creative Imaging
1.877.279.2441
  905.666.6647
________________________________________________________________________

 Date: Fri, 11 Apr 2003 18:07:30 -0400
   From: Jim Rich
Subject: Re: Offset lithograph dots

Howard,

Usually, when an informed person refers to halftone dots they are talking about positive dot values that reflect the percentage of an area, even if they are viewing halftone negatives.

As for the word "artifacts" produced by screening, do you mean something like stochastic that appear to be random? I have never heard of someone calling a halftone dot an artifact, but then, I have only been working with halftones for 35 years. So that does not mean that some new overnight sensation at a software house didn't come up with a new term for halftone screening.  

Jim Rich
________________________________________________________________________

Date: Fri, 11 Apr 2003 20:36:39 EDT
   From: Dan Margulis
Subject: Re: Offset lithograph dots

Howard writes,

When a commercial printer refers to "dots", is he (or she) referring to
halftone dots or to the artifacts produced by screening?  It seems that the
number of artifacts per linear inch is the same as the lpi, while the number
of halftone dots per linear inch appears to be--what?--eight times the lpi?

Halftone dots *are* what you are referring to as the artifacts of screening. I think what you are calling dots may be pixels--the smallest components of a digital file. Or possibly you are referring to the spots laid down by the platesetter, which together create the halftone dots.

The terminology is strange because the only item that the average person would refer to as a "dot" (the screen pattern) is referred to in terms of *lines* per inch, whereas pixels and platesetter spots are both commonly expressed in "dpi"--dots per inch.

Dan Margulis
 ________________________________________________________________________

Date: Fri, 11 Apr 2003 18:59:24 -0700
   From: Mac Townsend
Subject: Re: Offset lithograph dots

And the reason they are referred to in "Lines per inch," of course, is historic. Until as recently as the 60s halftones were made in a large camera using engraved glass "screens". These were assembeld from a pair of glass plates which had been engraved with straight lines so many lines to the inch. The engraved lines were then filled with carbon black (at least orginally, I don't know what the 110 lpi glass screen I have in the back shop is filled with). the two plates were arranged with the rules or lines at 90 degrees to each other. The effect if you used a loupe to look thru the assemblage was that of looking thru a common window screen. Thus the term "screen" is used to describe this. (FWIW, in the 60s glass screens were replaced by photographically produced plastic contact "screens". They no longer looked like a window screen when you used a loupe to look thru them, but the term had been in use for nearly a hundred years so it continues to be used today).

Mac Townsend
Adcom Graphics, Digital Imaging
Fairfield, California
www.adcomgraphics.com
________________________________________________________________________

Date: Sat, 12 Apr 2003 04:25:08 -0000
   From: "Howard"
Subject: spots, dots, and halftone dots

    Don't mean to be a nuisance about this, but I've been trying to find these answers for well over a year.  My terminology may be faulty in some cases, so let me put it in non-technical language and see if I've got it right this time.
    An imagesetter puts up to 256 black spots in a matrix representing the percentage of ink in one byte.  When the image is printed, sixteen of these matrices (halftone dots) are combined into a roughly circle shaped form which I perhaps mistakenly presumed to be artifacts produced by the screening process, though that's how I've seen them described in several books. Since the number of these circular collections of halftone dots is equal to the lpi, I wondered what name was used for them.  From what Dan Margulis replied, I presume these are just called "dots" as distinguished from "halftone dots".  Now, if that's the case, when a printer refers to "dots" he is apparently using the term as synonymous to "lines per inch".  But when he takes out his trusty loupe, examines a printed sample, and talks about "dots", is he looking at individual halftone dots, at the circular- or star-shaped groupings of sixteen halftone dots, or at both?  Most probably he is examining individual halftone dots to make a rough check on individual ink percentages in the area examined, but this is not exactly obvious to those of us with no prepress experience beyond approving proofs done by others.  
    Most of the authors whose books I've read--and there are many--  gloss over the whole thing as if it was too elementary to discuss.  Even a textbook on commercial offset printing managed to be extremely vague about the whole subject.  My impression has been that "a dot is a dot and don't worry about it".  But I do worry about it.
    I won't even ask about the dots produced by the piezoelectric method used by inkjet printers.  Everyone I ask has a completely different explanation.  All of them sound both authoritative and plausible, but the problem is that no two of them give the same explanation.  My confusion may well be due to my lack of knowledge of proper terminology but I feel like a person who asks for directions and is told by one to go north, another to go south, etc.
    If anyone still has enough patience to overlook my irritating persistence, I would appreciate any clarification you can offer.

I'm not sensitive to criticism, especially when I get something positive to go with it.  If either my terminology or my understanding is faulty, please don't hesitate to correct me.

Howard Smith
________________________________________________________________________

Date: Sat, 12 Apr 2003 11:10:19 -0400
   From: Jim Rich
Subject: Re: spots, dots, and halftone dots

 Howard,

I get a sense of what you are after, so if you can, why don1t you post some real pointed questions. That might make it clearer and easier to answer you.

Also, just to be sure it has been said: there are many terms and ways to describe resolution.

There is input resolution that has to do with cameras and scanners. This is usually   described in dots per inch and  pixels per inch. Though there are other measures.

Then there is output resolution that has to do with shades of gray per color and halftone screening. This is usually  described in spots per inch,  dots per inch and pixels per inch.  In some cases input and output resolution can be describing different measures, but the same words are used that can make understanding input and output resolution difficult.
     
Jim Rich
 ________________________________________________________________________

Date: Sat, 12 Apr 2003 11:16:16 -0500
   From: Ron Bean
Subject: Re: spots, dots, and halftone dots

"Howard"  writes:

An imagesetter puts up to 256 black spots in a matrix
representing the percentage of ink in one byte.  When the image is
printed, sixteen of these matrices (halftone dots) are combined into
a roughly circle shaped form which I perhaps mistakenly presumed to
be artifacts produced by the screening process, though that's how
I've seen them described in several books. Since the number of these
circular collections of halftone dots is equal to the lpi, I wondered
what name was used for them.

Are you talking about "rosettes"? These can only be seen when all four colors are printed together; individual plates (or films) don't have them.

If you have a copy of Yule's "Principles of Color Reproduction", there's a complete explanation of it (with all the math) in chapter 13. From page 339:

"If the screen angles in fig 13.07b were exactly 60 degrees, the moire pattern would become infinitely large. The dot arrangement would be similar over the whole sheet. It might be dot-centered of clear-centered, or it might have the irregular intermediate pattern. A shift in register of only half a screen period would shift it from one to the other. With the clear-centered pattern, there would be no point at which three dots were exactly superimposed; and with the dot-centered pattern, the centers of the clear openings would never coincide. The relative areas of the eight colored components of the pattern would change with this shift in register, and consequently the color would change slightly. This always occurs in four-color printing, especially when there is much black in the middletones, but it is usually not distinguished from other causes of color balance shifts. It becomes very noticable when an error in screen angle brings the different patterns close to one another in the form of moire."

He also says: "The arrangement of these rings and rosettes of dots is puzzling at first sight. They are not equally or regularly spaced nor equally symmetrical." So when you say that they're "equal to the lpi", I'm not sure if you're talking about the same thing.

[Yule's book is now published by GATF, or Amazon can get it for you-- it takes longer because they just order it from GATF, but some people find Amazon easier to deal with.]

The only other thing I can think of used to be called "supercells", which are a way of dealing with the fact that you're making variously angled screen patterns with a laser that always scans at the same angle. But you're not supposed to be able to see those.
________________________________________________________________________

Date: Sat, 12 Apr 2003 17:19:30 -0000
   From: "Roy Harrington"
Subject: Re: spots, dots, and halftone dots

Howard,

By far the best book I've read about this is:

Real World Scanning and Halftones
 by Blatner, Fleishman, and Roth.

It discusses halftones from image setters to stochastic dithering of inkjets.

Roy Harrington
www.harrington.com
________________________________________________________________________

Date: Sat, 12 Apr 2003 10:48:01 -0700
   From: "Mac Townsend"
Subject: Re: spots, dots, and halftone dots

What circular shapes are you referring to?

You won't see anything like this in a single color screening unless there are some problems with the image. In multi-color printing you will see something called the "rosette."

A screen has both a ruling or lines per inch (or frequency as it is called in computerese) characteristic AND an angular characteristic.

When multiple colors are printed on top of each other, as in a color photo, each of the 4 process colors is assigned a different angle (usually 105-75-90-45 for cmyk, but other vvalues are common, depending on factors not considered here).

This angular displacement of each color's screening is used  to prevent (or reduce the prominence of) interference patterns called "moire."

When the angles are correct and printed in register, the arrangement of dots slightly resembles a flower, thus a "rosette"--a similar pattern is achieved with just two colors as well, again printing at 30 degrees difference.

But the number and spacing of these rosettes is not the same as the lpi value of the screening or halftone dots themselves.

Mac Townsend
Adcom Graphics, Digital Imaging
Fairfield, California
www.adcomgraphics.com
________________________________________________________________________

Date: Sat, 12 Apr 2003 18:53:13 -0000
   From: "Howard"
Subject: Spots, dots, and halftone dots -- once again

    It was suggested that I rephrase my question to make it a little more to the point.  So let's try it again.

    An imagesetter employs up to 256 spots to create one halftone dot.

One halftone dot represents the percentage of color in one byte in an image file.  From what I can tell with a good loupe and several book illustrations, an image printed on an offset press consists of millions of almost microscopic rosettes.  Each rosette contains sixteen colored spots arranged in a roughly circular patter, and each of these spots of color is related to a halftone dot on one of the printing plates.  From what I can figure out, the number of rosettes per linear inch corresponds exactly to the lpi.  This makes sense because two ppi are used to obtain one lpi, and each rosette contains information representing four pixels.  (It took some time to reason out how you get four pixels in there with only two ppi.  Forgot that ppi are counted vertically as well as horizontally.)

     Before asking the following questions, I want to assure everyone that I know the difference between pixels, scanning samples, imagesetter spots, and halftone dots.  

     Now the specific questions:
     (1) Are the details listed above correct?  

     (2) And, just for information, p. 84 of Adobe's Print Publishing Guide calls an imagesetter spot a "printer dot (dpi)".  It shows a magnified view of a halftone dot and calls it a "halftone dot".  Then it shows a much tiner dot that I would swear represents the same kind of halftone dot except that the illustration labels it a "halftone cell".  Is a halftone cell the same thing as a halftone dot?

     (3) What are the printed rosettes called and what relationship do they have to lpi? (Or is there any relationship at all?)

     (4) How can the number of halftone dots per inch (dpi) equal the number of lpi?  Since each print color in a rosette is represented by four halftone dots, how can the numnber of halftone dots per inch equal the number of lpi if the number of lpi is the same as the number of rosettes per inch?  The key to this whole thing may well lie in the answer to this question.

     In case anyone wonders why I haven't approached local professionals in the commercial printing industry with these questions, I have.  Very interesting answers, but no two of them were the same.  Some simply looked confused and spoke in generalities.  All seemed to be sincere but just didn't know how to answer the questions.  Maybe my reasoning was so confused that I just wasn't asking the right questions, as one of you was helpful enough to point out.  It's something I hadn't considered, but it makes sense and I appreciate the suggestion.

Howard Smith
________________________________________________________________________

\Date: Sat, 12 Apr 2003 16:10:52 -0700
   From: "Mac Townsend"
Subject: Re: Spots, dots, and halftone dots -- once again

Are the details listed above correct?

1) I don't think so.

(A) The number of rosettes/inch is not equivalent to the lpi. The rosette is a construct of haftone dots. The size of halftone dots that make up the rosette is equivalent to the lpi. The size of the rosette will be geometrically related to the lpi, but cannot be the same figure and must be smaller. Why? because a garage that contains several cars cannot be the same size as one of the cars. Halftone dots, BTW, are physically measured insofar as lpi is concerned by theiur center-to-center distance, not their apparentl diameter. The apparent diameter is measured in percentage--equivalent to the percentage of the halftone cell that is filled.

(B) a digital halftone dot can and usually does represent more than one pixel of data. In some cases it represents less than one pixel (as when imaging a 72dpi jpg at 150 lpi). It has been suggested to me (in other forums when I asked how a halftone dot is made from the data presented to the screening engine) that certain areas within the halftone cell (within which the halftone dot is "grown") can be and are influenced by the data being converted to that area of the dot. Thus part of a dot might represent one pixel's value and another part a different value. I was told that the dot as a whole does not represent an average of the pixels being represented, but reflects some data from all the pixels it is to repressent. This fits in with Dan's observation/statement that an overabundance of data results in a "smoothing" effect in that the influence of discontinuities is reduced in the final rendering and that it is often these very discontinuities that we perceive as detail.

(C) PostScript level 3 permits many more than 256 imagesetter spots per halftone cell-dot. And supercell technology can string together cells into a single halftone dot.

each of these spots of color is related to a halftone dot

(D) These spots of color are not "related" to a halftone dot, they ARE  halftone dots. It is the rosetter that is "related" to the halftone dot.

And, just for information, p. 84 of Adobe's Print Publishing
Guide calls an imagesetter spot a "printer dot (dpi)".  It shows a
magnified view of a halftone dot and calls it a "halftone dot".  Then
it shows a much tiner dot that I would swear represents the same kind
of halftone dot except that the illustration labels it a "halftone
cell".  Is a halftone cell the same thing as a halftone dot?

The halftone cell is filled with printer spots. These printer spots are generated based upon the data in that that microscopic area in the image/page (it need not be a pixel, for pritner spots construct vector data as well). When the halftone cell is completed it becomes a halftone dot. The usual explanations show a nice symmetrical halftone dot growing from the center of the cell outward. It doesn't. Especially with modern supercell technology. A HT dot may be seeded at several locations within the single cell almost at the same time, within a supercell construct even more than a single cell.

How can the number of halftone dots per inch (dpi) equal the
number of lpi?

Because this is how the halftone dot is defined. It is measured in lines per inch, for historical reasons I outlined in a previous posting. The "conventional" way to produce halftone dots utilized a device called a screen which had cross-hatched lines and the coarseness/fineness of this screen was reckoned in the threads/lines per inch. Each little square in the screen generated a single halftone dot.

4) I might suggest that photoshop books will not lead you to a very useful understanding of this stuff. The halftoning process has been around for almost 150 years, Photoshop for perhaps a dozen.

You will find Blatner & Roth to use different terminology again. They refer to a dot as what the printer makes on the page, quite different from that used by others.

Such books as the Lithographer's Manual and others on the GATF bookshelf (www.gatf.org) could  also help you clarify you understanding of the printing aspects of the question.

Mac Townsend
Adcom Graphics, Digital Imaging
Fairfield, California
www.adcomgraphics.com
________________________________________________________________________

Date: Sat, 12 Apr 2003 20:56:36 -0400
   From: Jim Rich
Subject: Re: Spots, dots, and halftone dots -- once again

Howard,
   
One of the problems I think you are encountering is that there are two sides to what you want to know about. One side is about general halftone theory that has been used in the printing business since the 1879 and the more modern implementaion of it in PostScript based workflows where people discuss things like shades of gray and halftone cells. So that might explain why you seem to get so many answers. And they might all be correct, but then they might not.

And, just for information, p. 84 of Adobe's Print Publishing
Guide calls an imagesetter spot a "printer dot (dpi)".  It shows a
magnified view of a halftone dot and calls it a "halftone dot".  Then
it shows a much tiner dot that I would swear represents the same kind
of halftone dot except that the illustration labels it a "halftone
cell".  Is a halftone cell the same thing as a halftone dot?

It is my understanding that a halftone cell is the building block of creating a halftone dot in the PostScript world. So depending on the shades of gray you need in your image reproduction you choose how you want to create the halftone cell. This usually happens behind the scenes and in a lot of cases you only have the choice to pick a resolution mode in your imagesetter and then a halftone screen lpi.

   (3) What are the printed rosettes called and what relationship
do they have to lpi? (Or is there any relationship at all?)

One way to think about Rosettes is that they are the least objectionable moire pattern that you get when using angled screens to produce a color reproduction.

As a rule of thumb the finer the line screen ruling or LPI the harder it is to see the rossette patterns. Rossette patterns are a part of halftone color reproductions and if you are using traditional screening you cant get away from them.

   (4) How can the number of halftone dots per inch (dpi) equal the
number of lpi?  Since each print color in a rosette is represented by
four halftone dots, how can the numnber of halftone dots per inch
equal the number of lpi if the number of lpi is the same as the
number of rosettes per inch?  The key to this whole thing may well
lie in the answer to this question.

I am not quite following this question, but let me  say this. The terms halftone  screen ruling, lines per inch and halftone dots per inch are different terms for the same thing. As for rossettes, they will vary depending on the lpi you choose.
   
Jim Rich
________________________________________________________________________

Date: Mon, 14 Apr 2003 11:33:08 -0500
   From: David.Clark
Subject: Re: Howard Smith

Howard - -  you need a long out of print book "Adobe Accurate Screens" by Peter Fink.   It is the best written technical book I own on any subject. The most complicated concepts are made crystal clear.  Find a used copy on the web and do not loan it to anyone!

David
 ________________________________________________________________________

Date: Mon, 14 Apr 2003 18:00:15 -0000
   From: "Roy Harrington"
Subject: Re: Spots, dots, and halftone dots -- once again

You will find Blatner & Roth to use different terminology again.
They refer to a dot as what the printer makes on the page, quite
different from that used by others.

I think this is the terminology correspondences:

dot -- printer dot -- imagesetter dot -- dpi

spot -- halftone dot -- halftone cell -- screening -- lpi

-- the rosette is just a moire pattern artifact

pixel -- photoshop files -- multiple tones at one position -- ppi

--------------

Using Blatner's terminology:

A spot is very much like one pixel -- multiple tones at  one position.  However you never see a one-to-one correspondence between them.  There's usually a 1.5 to 2.0 ratio recommended.  This is not because the spot can somehow represent more than one pixel's information. It simply that the screening/halftone cells are usually done a various angles.  This means that an "effective resampling" of the input data occurs to get halftone cell info at positions that don't line up with the input data.  So its better to have more ppi in the input than lpi in the output to accommodate this resampling.

Roy Harrington
www.harrington.com
________________________________________________________________________

Date: Sat, 19 Apr 2003 23:28:54 -0500
   From: "Howard Smith"
Subject: Re: Spots, dots, and halftone dots -- once again

  From: Jim Rich
 
  I am not quite following this question, but let me  say this. The terms
  halftone  screen ruling, lines per inch and halftone dots per inch are
  different terms for the same thing. As for rossettes, they will vary
  depending on the lpi you choose.

      Slowly but surely this is beginning to make sense.  Your reply and those of several others have done more good than a year of reading books.  It helps to be able to consult with the experts.  But lt leaves me wondering how many years I have to keep living in order to reach that level myself.  But maybe no one really does since the learning curve seems to be without end.
  Sincerely,

   Howard Smith
______________________________________________________________________

Date: Sat, 19 Apr 2003 23:25:08 -0500
   From: "Howard Smith"
Subject: Re: Spots, dots, and halftone dots -- once again

  Mac,
      Thank you for a very complete and obviously authoritative answer to my question about halftone dots and lpi.  Trying to figure this out for myself has been one of the most frustrating experiences I've ever had.  Just when it looked like the answer was nice and clear-cut, it seems that it's even more complex than first imagined.   You also went a long way toward explaining why Photoshop books don't go very far into this kind of subject.  Apparently it takes a combination of experience along with attendance at a number of seminars, trade shows, etc. to begin to get a good grasp of some of the complexities.  Now it's easier to understand why Dan didn't explain all this in a couple of paragraphs in Professional Photoshop.  
      What a pity!  My conclusions were so nice and simple--and so totally wrong.  It leaves me almost fearful of delving farther into Photoshop, but then why not?  It's been quite an adventure so far.
  Sincerely,

  Howard Smith

Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.