Gradient Banding

RA
Posted By
Richard_Archer-Jones
Nov 21, 2006
Views
587
Replies
13
Status
Closed
I’ve created a document in PageMaker 7 and in the background I’ve placed a gradient which was produced in Photoshop CS2, but when it is printed out the gradient doesn’t always looked good, it’s got vertical banding and where it fades to white it just doesn’t do it smoothly. It looks alright on the screen though. In fact, it looks better on our in-house printers than from the professional printer I recently got a draft print out from. Why is this gradient not coming out right?

Initially the PS graphic wasn’t the right width and in PM it was stretched to fit, but now I’ve made it the right width in PS and I’m in the process of testing it at the professional printers (find out tomorrow). I’m wondering, what is the best way to produce gradients? I’ve read some of the postings I’ve found from a search on this subject but haven’t solved the problem because I’m not sure my problem was directly discussed, and some of the words used were beyond me. I understood something about 16 bits, but when I changed the PS graphic to 16 bits PM wouldn’t recognise it. I guess I’m going to get replies of "Get InDesign", but that’s easier said than done. Does ID take 16 bit graphics? Is there something I need to click in order to get PM to take 16 bit graphics?

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

H
Ho
Nov 21, 2006
16 bits won’t do your prints any good; the extra info has advantages when editing images, but none when it comes to output. What sort of device are you sending your output to? PS (postscript) Level 3 devices have less tendency to band than Level 2 since they support more gray levels.

Banding can be caused by using the wrong line screen, trying to stretch the gradient too far across the page and using colors that have the wrong relationship to each other.

Fading colors (or gradients) that stop abruptly can signal that the physical limits of your output device have been reached—printers can print dots down to a certain size or percentage, usually in the neighborhood of ~5%.

Generally, adding a little noise to the gradient can help with the banding, as can taking all of the above into consideration. Your printer (the man, not the machine) is the best resource to solve this problem. If he doesn’t know or care, find a new printer.

I would also suggest the Print Design Forum as the best place to get input from Photoshop/PageMaker/InDesign users.

<http://www.adobeforums.com/cgi-bin/webx?14@@.ef602c5>
RA
Richard_Archer-Jones
Nov 23, 2006
Hi Ho ….. ah I see, yes, very good … anyway, thanks for that. There’s stuff I don’t understand in your reply though i.e. "wrong screen line" and "colours that have the wrong relationship to each other".

I understand the noise bit. I don’t know the results properly from the professional printer yet but if what I did didn’t work I’ll try adding noise. I’ll take more notice of the Print Design Forum from now on. Thanks.
L
LenHewitt
Nov 23, 2006
Richard,

The linescreen is what breaks up an image into dots so tones can be printed using just 4 colours of ink, by printing larger dots for dense areas and smaller dots for lighter areas. The number of tones that a PostScript device can reproduce is depedent upon

1) The linescreen being applied and
2) The output resolution of the printer

For any given printer dpi, as the linescreen frequency increases, the number of reproducible grey tones decreases. Here is a ready-reckoner:Screen
frequency (lpi) Output device resolution (dpi)
150 300 400 600 1200 2540
30 26 101 179 401 1601 7169
40 15 57 101 226 901 4033
50 10 37 65 145 577 2582
53 9 33 58 129 514 2298
60 7 26 45 101 401 1793
65 6 22 39 86 342 1528
85 4 13 23 51 200 894
105 3 9 16 34 132 586
120 3 7 12 26 101 449
130 2 6 10 22 86 383
150 2 5 8 17 65 288
200 2 3 5 10 37 162
220 1 3 4 8 31 134
300 1 2 3 5 17 73

Intermediate values can be obtained by using the following formula:

(dpi/lpi)^2 = No. of available tones.

This does not take into account dot gain, tone splatter, PostScript limitations etc.

If you have the LJ4 set to 300ppi and a default linescreen of 53lpi (plc mode) then you will only get 33 distinct tones, which is not enough for even shading. At 600 dpi that will increase to 129 tones which will be much better.

But for what is normally accepted as ‘continuous tone’ (256 tones) you would have to drop the linescreen down to less than 40 lpi which would give a very coarse screen

Now professional output will use an imagesetter that is capable of at least 2500 dpi and could be as high as 5000+dpi. Linescreen frequencies generally used tend to be 150 or 175 lpi, although fine-art publications may use a higher line screen frequency.

Hope that helps…
B
Bernie
Nov 23, 2006
(dpi/lpi)^2 = No. of available tones.

Actually: (dpi/lpi)^2 + 1 = No. of available tones

(not that it makes a huge difference)
T
Tommo
Nov 24, 2006
..

Hi Richard:

Here’s what you do:

In Photoshop make your gradient with the gradient tool (not the gradient "effect"), and if the image is not to be scaled up, use 300ppi as the image resolution. Then, on the gradient only, go to Filter>Noise>Add Noise, and add 1% monochromatic noise. Hit Ok, then hit Command/Control F (to repeat noise filter) — a total of 3 times. If you intend to scale up the print to about twice actual size, use 600ppi document resolution when you make the gradient, or more if you are scaling up more. But if you use the higher PPI, then hit the noise filter more (eg. 300ppi will look good with about 3 hits of what I suggested – eg. monochromatic noise 1%, three times — but if you make the gradient at 600ppi then hit the noise filter about 5 times at these settings of 1% each time. This will give probably the closest possible to perfect gradients in high quality print.

(I don’t know much about Pagemaker but if your printer guy cannot get good results from your PS gradient, then save the PS gradient as a PDF, making sure interpolation is checked as you save, and that might work better for them.)

..
RA
Richard_Archer-Jones
Dec 7, 2006
Well, thanks very much for the extensive replies folks. Sorry I’ve been so long in replying but I appreciate your time and effort. Are you still there? I have spent sometime trying to understand the information and I have a few questions.

Len – I’m confused as to why the number of grey tones decreases as the linescreen frequency increases, especially when dot gain is not being taken into account. First, I’m puzzled why you specify grey tones and second I am under the impression that linescreen frequency is similar to dots per inch, and the more dots per inch in use I would have thought the more tones that are available.

Tommo – Thanks very much for those instructions (I’m using a Windows PC but I manged to figure it out). I see how it works. I will use that method in future. I’ve yet to find an ‘interpolation’ box to tick when making a PDF though.
B
Bernie
Dec 7, 2006
I’m confused as to why the number of grey tones decreases as the linescreen frequency increases

This is because the halftone dots in the linescreen (measured in lpi) are made up of printer dots (measured in dpi)

An example:

75 lpi screen on a 300 dpi printer: each halftone dot is made up of up to 16 printer dots. giving you a total of 17 halftone dots (zero printer dots, 1 printer dot…all the way to 16 printer dots which is all black) and so 17 shades of grey (including pure lack and pure white) are available

150 lpi screen on a 300 dpi printer: each halftone dot is made up of up to 4 printer dots. giving you a total of 5 halftone dots (zero printer dots, 1 printer dot, 2 printer dots, 3 prionter dots and 4 printer dots, which is which is all black) and so 5 shades of grey (including pure lack and pure white) are available
RA
Richard_Archer-Jones
Dec 7, 2006
Mmmmm, thanks Cybernetic. I vaguely see what you’re getting at but I can’t get my head around all these dots. How can each halftone dot be made up of more than 4 dots?

This is my understanding of the situation. For colour printing there are specific patterns of dots for each colour, or rather the pattern of dots are at a specific angle on the paper. There are 4 colours, cyan, magenta, yellow, and black, and each colour has a pattern of evenly spaced dots at a specific angle set so that when all the dots for each colour are placed on top of each other they create an even tone (continous tone) without any moire patterns.

Therefore it depends what the definition of a halftone dot is. My interpretation is that for black and white printing a halftone dot is just one dot which will either be black or white and if it’s black it will differ in size depending on type of grey or black required. For colour printing a halftone dot is composed of four dots, one for each colour and each dot differs in size depending on what colour is required.

Do you see my problem? I can’t see what is meant by printer dots and halftone dots, and how a halftone dot can be made up of any more than 4 printer dots.
B
Bernie
Dec 7, 2006
Consider the following:

1) Each printer dot has to be the same size

2) Halftone dots must vary in size to convey varying shades of grey

3) Thus in order to make variable size halftone dots, one makes them out of smaller printer dots and varies the number of printer dots.

Maybe this will help to visualize it:

< http://www.pixentral.com/show.php?picture=1UFbZ7kvlWjt4D685Q NweEb9rwDv5>

From this, the math oriented folks can calculate the number of avaivable shades with the following forumla:

Shades = (lpi/dpi)^2 + 1
L
LenHewitt
Dec 7, 2006
Hi Richard,

First, I’m puzzled why you specify grey tones<<

When we talk about commercial printing, that usually means a conventional press using CMYK process inks, and offset Litho printing.

That requires the image to separated into those four colours – just like the channels in a CMYK Photoshop document. Each of those channels is a greyscale channel, and the separations derived from a CMYK document are likewise greyscale. Thus we have to consider each separation (channel) as a seperate entity and those are greyscale.

Now, when the file goes to the RIP, the output is sent to an imagesetter that images the output onto very high contrast film – Lith Film – which can only produce dense black and clear film (when exposed and processed correctly).

In order to record tones we have turn those tones into different sized dots, as we can only record dense black and clear film.

I am under the impression that linescreen frequency is similar to
dots per inch, and the more dots per inch in use I would have thought the more tones that are available.<<

As we are producing a film negative (some systems will create film positives rather than negatives), the lighter the tone, the larger the dot has to be and vice-versa. This is known as a linescreen.

Now the imagesetter can have a resolution in excess of 5800 dpi and will certainly have a resolution of at least 2540 dpi.

The linescreen, though, will have a screen frequency of between 80 and 200 lines per inch.

Now, the more ‘printer dots’ (controlled by the dpi of the imagesetter) we have to make one completely black linescreen dot, the more different tones we can record in that linescreen dot. So the higher the imagesetter dpi and the coarser the linescreen dot, the more tones we can record. If we use a finer linescreen, then the linescreen dots become smaller, and it takes less imagesetter dots to completely fill one of those linescreen dots, and so less variation in the number of imagesetter dots we can place in each.

Consider a linescreen of 200 lpi and an imagesetter of 2000dpi.

Each linescreen dot will be completely filled by a 10 x 10 block of printer dots. Now, the maximum theoretical range of that dot is from 100 to 0 , or a 101 separate tones.

If we increase the linescreen frequency to 400 lpi, then it only takes a 5 x 5 block of printer dots to fill one linescreen dot and the maximum possible number of tones will be 5 x 5 +1 or 26 tones.

Sorry for the long drawn out explanation, but I hope that helps.
RA
Richard_Archer-Jones
Dec 14, 2006
Ah, thanks Cybernetic, I see now. Halftone dots are composed of printer dots.

Gosh, thanks very much Len. I understand why grey scale is referred to. It’s like computer language (binary) isn’t it? The basics of computer language is ‘off’ and ‘on’, and the basics of printing language is ‘black’ and ‘white’ (or ‘colour’ and ‘no colour’). I’m struggling with the rest of your helpfull message though. I’m trying to get stuff done before Christmas but I’ll have another go at studying it next week. Thank you folks.
LJ
Leona_J
Apr 8, 2007
Is CS3 ging to make this problem any easier to deal with ???
LH
Lawrence_Hudetz
Apr 8, 2007
Well, well, here’s Len and Phos, OT on a thread! Who’d imagine! 😀

Happy Easter!

Must-have mockup pack for every graphic designer 🔥🔥🔥

Easy-to-use drag-n-drop Photoshop scene creator with more than 2800 items.

Related Discussion Topics

Nice and short text about related topics in discussion sections