2

in offset printing, for example, the cmyk colors are printed as dots of different sizes. One of the questions for me is how many size options are there for the individual colored dots?

The reason for the question is that Photoshop, for example, works with values ​​from 0-100%, but saves the colors in the color channels in 256 possibilities if you are in 8-bit mode.

As far as I know, the laser exposes with 8 bits in offset printing. are the binary and percentage values ​​in the print both correct?

3
  • 1
    As a retired offset litho machine op, the answer is actually 'infinite' because you got about 50 manual screw taps on the front of each ink dock, which you can twiddle to your heart's content. You can even change your ink transfer density slightly by changing the plate dampening.
    – Tetsujin
    Jul 23, 2022 at 13:45
  • I must add that with offset litho you also have to consider how much ink is in each dock 'stripe', ie the amount of ink released by each dock key. The rollers beneath the dock have lateral motion to even out ink flow, so one key does not affect only one 'stripe' by the time it reaches the plate. This means if you have a heavy area next to a very light one, you are going to get 'cross-talk' where each 'stripe' is influenced by the weight of ink near it. This must be taken into consideration when making your layouts. […cont'd]
    – Tetsujin
    Jul 23, 2022 at 18:02
  • [cont'd…] A good traditional shop will advise you on this before making plates, but they'll attempt to print anything you ask for if you choose to ignore their advice. They may make you pay up front, though ;)
    – Tetsujin
    Jul 23, 2022 at 18:04

2 Answers 2

2

Most modern prints are digital in origin. There are several units to consider.

First the file. As you know you can have different bit depths on a file. Current systems use 8 bits per channel (you already knew that).

There is an additional unit in the mix besides bit depth, LPI, this is, Lines per Inch. The amount of shades a plate can provide depends on the relation between the defined LPI and the DPI (Dots per Inch) the device can handle.

If you need 150 LPI, and your device can print 2400 DPI, then each line has at its disposal 16 linear dots (2400DPI/150LPI = 16 dots)

This means you have at your disposal a square grid of 256 potential combinations of dots to produce an LPI dot.

If you vary the relationship between the desired LPI and the DPI the printer is capable of, you can have a different amounts of shades any given grid can provide.

enter image description here

Any modern plate printer can give you around 3600 DPI, which means that you can push the LPI resolution to 200LPI (depending on other factors such as the type of paper you are using)


The Scale of 0-100% is due to historical reasons, before the digital era. But it gives you enough room to make a series of swatches. But if you are preparing a gradient, this gradient does not jump from 50% to 51% for example. It is translated on the background to its equivalent to, let's say an 8-bit scale. You can see this if you export the file to a raster image.

It is not necessary to use more values than a 0-100% scale, because the visual difference is so insignificant, and the printed result has so many additional variations that a small dot of those 256 possible dots is not a big deal.

The real dot size depends on the absorption of the paper, the gain, the pressure of the rolls, the humidity levels, ink density, etc. A tiny dot of the 256 available is not that important.


There are some additional elements to consider. The grid I am showing is not as direct in real life, because these grids and halftones are rotated, so the final dots used on each grid vary. Also, you have the shape of the "circle" that can be an ellipse, a diamond, etc.

But is, IMHO a good explanation to understand the number of shades on a print.

You can also have the scholastic method to produce shades, this is similar to the one used on an inkjet printer, using an error diffusion to distribute the dots to produce shades.

1

This is a configurable thing, there is no reason the raster size couldnt be what you want it to be.

But usually its a bit less than 256 levels as that would make the raster at minimum 16 dots wide. Which is fine if your doing stochastic rasterisation. If not it might be wider than that. Anyway that would mean that a 600 dpi printer would have a 37.5 ppi resolution which is a bit low even if assuming that the raster could be weighted still not great at about 75. Dropping it to 100 levels would give you 60 and 120 respectively thats slightly more acceptable.

There is also no reason why you must use the drivers rasterisation engine either.

2
  • So the color depth is only related to the raster size and has nothing to do with the binary system or with percentages? why is cmyk given as a percentage at all?
    – Alex
    Jul 23, 2022 at 14:00
  • @Alex because typically thats the resolution of color you can expect to get. Anyway the system underlying is not binary at all times so you can have more colors if you really want to.
    – joojaa
    Jul 23, 2022 at 14:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.