Round vs Euclidean

Ok, that makes sense.
In poking through my settings I see that.
and I am working with pre-screened 1-bit tif files. at least, that is what the Xenith sends to my plate burner.

so, if I'm getting all this right, you're saying that a round dot with standard screen angles would be better than the Euclidean with the modified screen angles?
 
Ok,and I am working with pre-screened 1-bit tif files. at least, that is what the Xenith sends to my plate burner.
so, if I'm getting all this right, you're saying that a round dot with standard screen angles would be better than the Euclidean with the modified screen angles?

Just to be clear, when I said: "The difference between 2540 to 2400 (2540dpi/100dpmm, 2400 dpi/96dpmm/) should not be significant unless you are working with prescreened (1 bit) files." what I was referring to was your customers sending you prescreened files. The plate burner will always receive 1 bit screened files from your RIP. However, if you receive a file from your customer that is 1 bit (already screened) then you will have artifacts if the prescreened file's resolution is not an even divisor of your output device's resolution.
Here's a comparison:
RoundvsEuclidean.jpg

The upper screen is Euclidean.
The lower screen is Round dot.
They are very similar in that both use round dots for the majority of the tone scale.
With a Euclidean the 51%-->99% dots are just a negative of the 1%-->49%. At 50% the dots convert to a square. That's why it is sometimes called a "transforming" or "composed" dot.
With a Round the dots from 1%-->99% are the same, they just grow in size.
As a general rule Round dots are most appropriate for a CTP work flow - and for reasons already listed are the best AM screen dot shape. There is much less, if any, need to offset the screen angles by 7 or 7.5 degrees to eliminate single channel moiré.
If your RIP allows for a Round dot, you should try it. You may (or may not) need to put a curve in the plate if you lose some shadow detail.

hope this helps, gordo

BTW, don't examine my Round dot gradient to critically - it is for illustration purposes only. I don't have access to a RIP so I had to create the round dot threshold array and use PShop as a RIP - so the dots are not as perfect as they would be in real life.
 
Last edited:
We were having problems with single color gradients, along with gradients that had up to 4 process colors in them using a round dot. Just this past week we switched to a euclidean dot, and they all came out beautiful. The round dot was fine for us as long as we didn't have any gradients on the job, but when we did there always seemed to be a problem with banding/stepping. The euclidean took care of that, and gave a very soft transition from top to bottom at 175 line screen.
 
We were having problems with single color gradients, along with gradients that had up to 4 process colors in them using a round dot. Just this past week we switched to a euclidean dot, and they all came out beautiful. The round dot was fine for us as long as we didn't have any gradients on the job, but when we did there always seemed to be a problem with banding/stepping. The euclidean took care of that, and gave a very soft transition from top to bottom at 175 line screen.
Interesting.
There is nothing about a round dot screen that would cause banding that wouldn't also appear with euclidean. Round dot screens also excel at vignettes. Effect does not always determine cause.

best, gordo
 
Last edited:
We were having problems with single color gradients, along with gradients that had up to 4 process colors in them using a round dot. Just this past week we switched to a euclidean dot, and they all came out beautiful. The round dot was fine for us as long as we didn't have any gradients on the job, but when we did there always seemed to be a problem with banding/stepping. The euclidean took care of that, and gave a very soft transition from top to bottom at 175 line screen.

Perhaps these results are related to the tone bump characteristics for each dot shape that Gordo listed in his first post? I am assuming that nothing other than dot shape has changed for the exact same job printed with both dot shapes.

4) The tone bump that occurs when dots touch occurs in the shadows at 75% so is much less visible in presswork (e.g. vignettes/skin tones) compared to Euclidean (50% bump where dots form a checkerboard) or Elliptical 40% & 60% bump.

The tone bump associated with a particular dot shape should be incorporated into the calibration curve applied to the plate exposure. Data that is typically derived from some manner of press fingerprinting or profiling.

If you change the dot shape you are in effect invalidating your press fingerprint or profile data.

In essence, by changing your dot shape from Round to Euclidean, you likely opened up your 3/4 tone range by several percent, thereby increasing the shadow contrast. You likely also increased the contrast from the highlights to 1/4 tone range. Lastly the mid to 3/4 tone range would likely lose some contrast. Is this what you observed?

What I am suggesting is that any time you make a significant change to your press, you should re-profile/fingerprint/calibrate it. The tone bump changes associated with a dot shape will change the gradient curve in ways that other press variables often can't. In short, dot shape is part and parcel with the plate-setter calibration curve itself.

I personally always wanted to know precisely why something worked. This is the best I can think of as to why the dot shape change to Euclidean worked for you. It also implies that the calibration/profile that you were using to image your round dot plates were likely less than ideal by the same token.

Best Regards
otherthoughts
 
Last edited:
Sorry all there is no such thing as a "best" dot shape, all work pretty well for that what they where designed for, Round, Euclidian, Elliptical, AM, FM.
Not only that different companies use different spot functions to create their halftone dots. That means that AGFA's round or euclidian dot is not going to be the same as that from Harlequin, It all boils down to what you want to print. For instance something with a lot of pastel tones or skin tones would be a nightmare to print in Euclidian because of the somewhat abrupt cutoff in the mid-tones, elliptical would be better. A round dot has it's problems in the area of 70% to 80% tones, if your prints are dark above the halftones try round the jump is in the dark area anyway, A limitation would be printing with a heavy ink coating (screenprinting) of over 300% the image will be swamped over the three quarter tones. At the moment I am in love with mixing CMK in AM round dot and Y in FM coarse (or the least color in FM) this way I have a nice reach over the entire tonal range, it also keeps artefacts from forming.
 
Sorry if slightly OT but 2400 vs 2540 and screening

Sorry if slightly OT but 2400 vs 2540 and screening

Interesting discussion on dot shape and someone brought up the issue of 2540 vs 2400. My question is what happens when the front end RIPs at 2400 but the native resolution of the device is 2540? In our instance ApogeeX going to a Magnus 400. What method(s) are used to change resolution of the device to image at the lower resolution?

We are having a lot of variability in dot size and I'm wondering if there is any connection. All my previous experience with CTP was with violet/silver plates, but I was told that thermal was much more stable. This has not been my experience with thermal so far.

Clark
 
Slammer wrote: "Sorry all there is no such thing as a "best" dot shape, all work pretty well for that what they where designed for, Round, Euclidian, Elliptical, AM, FM."

GP: As as far as AM and CTP for offset is concerned - round dot is the best - see my original post for the reasons.

Slammer wrote: "Not only that different companies use different spot functions to create their halftone dots. That means that AGFA's round or euclidian dot is not going to be the same as that from Harlequin"

GP: Because the round dot is just that - unlike other spot functions - there is no significant difference between the different vendor's implementations - another benefit of the round dot.

Slammer wrote: "For instance something with a lot of pastel tones or skin tones would be a nightmare to print in Euclidian because of the somewhat abrupt cutoff in the mid-tones, elliptical would be better. A round dot has it's problems in the area of 70% to 80% tones, if your prints are dark above the halftones try round the jump is in the dark area anyway"

GP: Precisely why you shouldn't bother with Euclidean or Elliptical as they offer no value over the round dot and they have issues in important tone areas, i.e. 1%-75%

Slammer wrote: "At the moment I am in love with mixing CMK in AM round dot and Y in FM coarse (or the least color in FM) this way I have a nice reach over the entire tonal range, it also keeps artefacts from forming."

GP: Glad you do use the round dot. Y printing in FM is a useful way to avoid inter-screen moiré. It can, of course, be used with CMK that's using any AM dot shape.

best, gordon p
 
Last edited:
Interesting discussion on dot shape and someone brought up the issue of 2540 vs 2400. My question is what happens when the front end RIPs at 2400 but the native resolution of the device is 2540? In our instance ApogeeX going to a Magnus 400. What method(s) are used to change resolution of the device to image at the lower resolution? We are having a lot of variability in dot size and I'm wondering if there is any connection. All my previous experience with CTP was with violet/silver plates, but I was told that thermal was much more stable. This has not been my experience with thermal so far.

You may get artifacts, however, because you are going to a higher res plotter (2400 --> 2540) you probably won't have any issues. Going the other way, i.e. 2540-->2400 will cause artifacts.
The usual strategy is to resample the bitmap in a 2540 --> 2400 dpi situation.

All things being equal, thermal should deliver more consistent dots on plate than visible (depends on many factors though). You should contact your CTP vendor (loudly) if you feel you are not achieving what was promised or expectations were not met with performance.

best, gordon p
 
Last edited:
GP: Precisely why you shouldn't bother with Euclidean or Elliptical as they offer no value over the round dot and they have issues in important tone areas, i.e. 1%-75%

In my personal experience, I ran Heidelberg's IS Classic, Elliptical dot at 2540/175 & 200 on coated stock for years and years and never had one single moire in any flesh tones or the likes. We ran the following screen angles C-165 M-45 Y-0 K-105. Maybe I was just lucky, doubt it tho.
 
In my personal experience, I ran Heidelberg's IS Classic, Elliptical dot at 2540/175 & 200 on coated stock for years and years and never had one single moire in any flesh tones or the likes. We ran the following screen angles C-165 M-45 Y-0 K-105. Maybe I was just lucky, doubt it tho.

We also ran euclidean for years..C165, M105, Y60, K45. Lots of pastels, no moire issues that I can recall. Looking at some old press tests using both round and euclidean, the tone bumps for either are essentially invisible to my naked eye. Just my experience...
 
In my personal experience, I ran Heidelberg's IS Classic, Elliptical dot at 2540/175 & 200 on coated stock for years and years and never had one single moire in any flesh tones or the likes. We ran the following screen angles C-165 M-45 Y-0 K-105. Maybe I was just lucky, doubt it tho.

I didn't say anything about screen angle moiré (i.e. moiré due to screen angles conflicts between process colors). I mentioned single channel moiré which is a different issue.
BTW, Running black at 105 gets rid of the cyan/yellow moiré (making it a K/Y moiré instead which is less visible) However K at 105 degrees is not as good as K at 45 degrees in monochrome work (e.g. B&W printing) because the screen dots are more visible at 0 and 90 degrees and least visible at 45 degrees. That is why Y (lightest color) is traditionally placed at 0/90 and black (darkest color) at 45. Most reference books on screening will have a graphic that shows this optical effect. It's more apparent with coarser screens so you might not notice it at the 175/200 lpi you're running).

best, gordon p
 
Last edited:
I didn't say anything about screen angle moiré (i.e. moiré due to screen angles conflicts between process colors). I mentioned single channel moiré which is a different issue.
BTW, Running black at 105 gets rid of the cyan/yellow moiré (making it a K/Y moiré instead which is less visible) However K at 105 degrees is not as good as K at 45 degrees in monochrome work (e.g. B&W printing) because the screen dots are more visible at 0 and 90 degrees and least visible at 45 degrees. That is why Y (lightest color) is traditionally placed at 0/90 and black (darkest color) at 45. Most reference books on screening will have a graphic that shows this optical effect. It's more apparent with coarser screens so you might not notice it at the 175/200 lpi you're running).

best, gordon p
my print blog here: Quality In Print

Which screen angle do you prefer between 105-45-90-75 and 105-75-0-45?
Thanks.
 
Regarding the monotone printing, I have set up different virtual printers (that's what metadimensions calls a print que basically) for different types of printing at the RIP.
For example, CMYK, CMYK + Spot, Spot Color work (including duotones), Black & White. Thus the B&W work isn't printed at 105, but 45.
 
Which screen angle do you prefer between 105-45-90-75 and 105-75-0-45?

In your example: 105-45-90-75 and 105-75-0-45 - both 90 and 0 are the same angle. Note also that 105 is the same screen angle as 15.
So, 105-75-0-45 is the same as 15, 75, 0, 45 which is pretty much the standard screen angle set used in N.America. Your other screen set: 105-45-90-75 - same as 15, 45, 0, 75 is the one typically used for caucasian skin color (avoids M/Y conflict but introduces C/Y conflict). So they are both valid depending on what colors are predominant in your image(s).

Here's a bit more info.

The following are typical screen angle sets. Note that interscreen moiré becomes more visible when the angles of any two screens are less than 30 degrees apart. Yellow is usually allowed to be less than 30 degrees because it is such a light color, the moiré is not usually visible. Also, the frequency of the yellow printer is usually made higher than the other three colors (typically around 108% higher) to further minimize the visibility of the moiré. The moiré can become more visible if the yellow printer becomes contaminated/dirtied by the preceding process colors, or if its density is too high. Some vendors offset these angles by around 7.5 degrees. That is a technique to minimize single channel/color moiré - not moiré in two+ color situations.

The following are all valid and generally used screen angle sets. Note which two are less than 30 degrees apart, if the image is mostly made up of those two colors then that screen set should be avoided for that image if possible.

The sequence for these screen sets is C, M, Y, K (i.e. the first screen set on the list is: 15C, 45M, 0Y, 75K). Screen angles have quadratic symmetry so 0 degrees is the same as 90, 180, and 270 degrees.

Standard 4/C U.S.
15, 75, 0, 45

Other screen angle sets

15, 45, 0, 75
15, 75, 0, 45
15, 45, 30, 45
45, 15, 0, 75
45, 75, 0, 15
75, 15, 0, 45
75, 45, 0, 15
75, 15, 60, 45

For 2/C
Dark at 45
Light at 75

For 3/C
Dark at 75
Medium 45
Light at 15

Skin color (avoids M/Y conflict but introduces C/Y conflict)
15, 45, 0, 75

Light greens (avoids C/Y conflict but introduces M/K conflict))
45, 75, 0, 15

best, gordon p
 
Last edited:
great reading here - very interesting discussion about round and euclidean. My company is currently going through a similar situation. We have been running Euclidean for some time (i am 3 years at this shop, and it has been Euclidean since i arrived), but i have some doubts about it, particularly in the flesh tones and in gradiants, as mentioned earlier in the thread.

my question though is in regards to linearization of plates. We have a plate linearization curve that we have applied to ensure that my 50 reads a 50 with the LithoCam. well, the good news is that it works. The 50 is indeed reading as such. My confusion is with what the 50 LOOKS like in a Euclidean pattern. What i have always read and learned and understood about the Euclidean pattern is that at the 50, you should have the perfect checkerboard pattern. What i am seeing does not agree with this. The pattern i see is definitely reading a 50, but it does not view as expected. See the attachment.

i contacted Kodak and Ihara and discussed this topic with them, and both seemed satisfied that my 50 is indeed a 50 and that the look shouldnt bother me, but it still does, because i am not understanding the reason why. we are using a1-Bit tiff workflow, outputting tiffs at 2540 from Prinergy, 175 LS, Euclidean dot shape to a Lotem Quantum 800 exposing at 2540.

I took the screen capture and brought it into photoshop and verified that it is indeed measuring 50% coverage, and all of the other dot sizes are measuring properly on the plate. It is just that it isn't giving me that picture perfect checkerboard that is puzzling me.

Any thoughts or ideas on this would be very well appreciated.
 

Attachments

  • bitmap 50.JPG
    bitmap 50.JPG
    53 KB · Views: 264
[SNIP]my question though is in regards to linearization of plates. We have a plate linearization curve that we have applied to ensure that my 50 reads a 50 with the LithoCam. well, the good news is that it works. The 50 is indeed reading as such. My confusion is with what the 50 LOOKS like in a Euclidean pattern. What i have always read and learned and understood about the Euclidean pattern is that at the 50, you should have the perfect checkerboard pattern. What i am seeing does not agree with this. The pattern i see is definitely reading a 50, but it does not view as expected.
i contacted Kodak and Ihara and discussed this topic with them, and both seemed satisfied that my 50 is indeed a 50 and that the look shouldnt bother me, but it still does, because i am not understanding the reason why. we are using a1-Bit tiff workflow, outputting tiffs at 2540 from Prinergy, 175 LS, Euclidean dot shape to a Lotem Quantum 800 exposing at 2540.
I took the screen capture and brought it into photoshop and verified that it is indeed measuring 50% coverage, and all of the other dot sizes are measuring properly on the plate. It is just that it isn't giving me that picture perfect checkerboard that is puzzling me.

I'm speculating here, but, despite what Kodak and Ihara says, I don't believe the image you posted is of a 50% dot. I think that it is more likely a 53%-55% dot.

Here's why.

Below is an image of a Euclidean AM dot direct from a Prinergy RIP:

3570.jpg


It is a section from a gradient. You can clearly see that at a certain point a checkerboard is formed.
So how do you confirm that the checkerboard represents a 50% tone?
One way is to import the bitmap into PShop, convert it to greyscale and slowly apply the Gaussian blur filter until you cannot distinguish the halftone dots anymore. I've done that to the below image. The top half is blurred and the bottom half is the original bitmap:

5053.jpg


If you then use the "Get Info" tool in PShop it will report that the checkerboard represents a 50% tone - which makes sense. However, a little to the left is a dot structure that looks very much like your microphoto. Measure that and PShop reports that it is actually about a 54% tone.

So I then took the image you provided and converted it to a bitmap and performed the same blurring. I also oriented the dots to the same screen angle:

Microphotocompared.jpg


Guess what? PShop reports that your original bitmap represents a 53% tone.

So, why the discrepancy? Well one reason might be in the way instruments calculate dot area from a microphotograph of a halftone on plate.
Most, if not all, use a thresholding algorithm to determine what is the non-printing plate and what is the printing dot. Put another way, the software decides that a pixel of X tone level and lighter is the plate while levels darker are ink carrying dots. Because the photo and/or dots have a slight softness to them, the result is a slight ambiguity as to where the transition occurs. Change the threshold and you change the size of the dot the instrument "sees" and hence the tone it represents. Below is your image on the left:

Threshold.jpg


and beside it are the same dots but sliced by a different threshold value. As you can see, changing the threshold indeed changes the tone value reported by the instrument.

My suggestion to you is do a vignette of the black printer and see if you achieve a checkerboard at the 50% tone. If you have the appropriate software you might also look at the original bitmaps before they are imaged to plate. You could also create a synthetic 50% checkerboard in Illustrator or PShop, image that and measure with your instruments to see if they report a checkerboard as being a 50% tone.

BTW, since 175 lpi is not an even divisor of 2540 dpi you will never get a perfect checkerboard - but you will get very close with just a couple of extra or missing pixels.

BTW, how did you confirm the tone value using PShop?

Hope this helps. Best gordo
 
Last edited:
wow, great response, and i have some thoughts id like to add. i just have a storm of work around me today, and dont have much time. I didnt want to seem like an ingrate by not responding right away! when i free up some time i will give you a better response.

thanks!
 

PressWise

A 30-day Fix for Managed Chaos

As any print professional knows, printing can be managed chaos. Software that solves multiple problems and provides measurable and monetizable value has a direct impact on the bottom-line.

“We reduced order entry costs by about 40%.” Significant savings in a shop that turns about 500 jobs a month.


Learn how…….

   
Back
Top