Why not line pairs per millimeter?

[Marc Wossner]

I wonder why I can´t find such a traditional resolution measure like
line pairs per millimeter when the capabilities of printers are
defined. All that´s stated is the dots per inch value. Is the lp/mm
figure meaningless when inkjet- or laser printers are concerned?
Furthermore I read that all printer drivers resample the image data to
the printers native resolution (720ppi for Epson and 600ppi for
Canon). If that´s true (and I´m not convinced it is because I also
found statements saying that inkjet printers use frequency modulated
rasterization and so line screen rulings do not aply) an Epson should
be capable of producing 720/25,4/2=14 lp/mm. That´s far more than the
6,88 lp/mm a human with average vision can resolve. Is this the reason
why manufacturers don´t state this value: It´s beyond our
capabilitites? I´d be happy if someone with deeper knowledge of the
digital printing process shares his/hers thoughts!

Marc Wossner

 

[Arthur Entlich]

I may not have this completely correct myself, and I'm running on little
sleep which always makes my theory recall poorer, but I believe line
pairs are a measurement of actual optically measurable resolution, while
dots per inch is a mechanical capability of the printer in terms of
positionable capability for any one dot. In other words, the dpi of the
printer output doesn't necessarily translate into a true measurable
resolution.

Further, there are two types of dpi to speak about in inkjet printers.
The actual potential output of dot positions on the paper, and the print
driver rasterization within the spooler. So, in the case of an Epson
printer, the printer, may be capable of printing a dot within a matrix
of 2880 dpi, or even higher, while the rasterized image within the
driver will only be resolved to 360, 720 or 1440 dpi depending upon the
model (the wide carriage Epson's for instance usually use one half the
driver rasterization resolution of the same vintage desktops). Also,
keep in mind that the ppi (pixel per inch, which refers to the smallest
discrete size of any one color that might appear on the screen, and
output dpi are not the same, since inkjet printers use a matrix of
numerous ink dots to build a pixel color.

Most inkjet printers use FM or another modulation system rather than
discrete cells to make a color from the dots, which gives better
transitions, and can be accomplished because the printers produce the
print progressively.

I don't know if this helps to answer you question fully.


Art

 

[frederick]

I wonder why I can´t find such a traditional resolution measure like
line pairs per millimeter when the capabilities of printers are
defined. All that´s stated is the dots per inch value. Is the lp/mm
figure meaningless when inkjet- or laser printers are concerned?
Furthermore I read that all printer drivers resample the image data to
the printers native resolution (720ppi for Epson and 600ppi for
Canon). If that´s true (and I´m not convinced it is because I also
found statements saying that inkjet printers use frequency modulated
rasterization and so line screen rulings do not aply) an Epson should
be capable of producing 720/25,4/2=14 lp/mm. That´s far more than the
6,88 lp/mm a human with average vision can resolve. Is this the reason
why manufacturers don´t state this value: It´s beyond our
capabilitites? I´d be happy if someone with deeper knowledge of the
digital printing process shares his/hers thoughts!

Marc Wossner

Line pairs printed using one colour will resolve better than the same
line pairs printed using several different colours. So with CYMK inks,
while maximum resolution might be achieved with one colour used, when
say green requires yellow and cyan, then the two coloured dots are
placed alongside each other - not blended together - the result may be
different. The droplets are also going to bleed or "splat" out when
they hit the paper - to different degrees on different papers. So the
theoretical maximum is not really possible.
There's an article / review here of using an Epson pigment printer
(R800) to produce high resolution prints for magnified "stereo card"
viewing. They conclude that maximum resolution that could be achieved
was around 10 lp/mm - but note that they did not feed at native 720 dpi.
http://www.crystalcanyons.net/pages/TechNotes/R800Printer.shtm
With a similar printer fed at 720 dpi, I could clearly resolve up to
about 10 lp/mm - about the same in the vertical direction - but a little
less in the horizontal direction or angled (lines resolved better in
direction that the paper feeds through the printer). Under a
microscope, the individual droplets are elongated oval in that direction
- not circular. With "RPM mode" (so called "optimised" 5760 x 2880)
there was a visible difference in dot placement, but not actual
resolution as far as I could see - but RPM mode looks better in terms of
colour tonality in shadow areas.
Whether there really is a noticeable practical advantage in resampling a
photo (as opposed to printing with test charts created at different dpi)
an image to (for Epson - 360dpi) before printing, or whether the printer
driver does a better job of this is a moot point. My opinion is that
it's probably better to do it yourself - resample (ie uprez) an
unsharpened image, carefully apply sharpening, then print.

 

[Marc Wossner]

Thanks for your replays.
So to check my understanding of the case: The upper limit of the true
resolution of an inkjet printer is given by it´s rasterization like
720 ppi/25,4 mm/2=14 lp/mm. But it can only come close to that if it
prints with one colour. If more than one colour is necessary to print
a tone it has to sacrifice resolution for tonal rendition because this
colour is formed by several dots alongside each other. So the true
colour resolution can not be calculated but must be measured by means
of a real print. So the reason for the manufacturers not stating lp/mm
values might well be that, due to the dependency on the different
colours, there is no single value like this. A laser printer like the
Lightjet 5000 on the other hand could state a single number like this,
because it produces real halftone pixels (305 ppi or 406 ppi) that can
be related to lp/mm. - Any truth in there?

Marc Wossner

 

[Arthur Entlich]

Hi Marc,

As I understand it, the Lightjet 5000 uses laser light to illuminate
photo paper. It therefore uses a continuous tone rather than halftone
cell or screen method to produce the image. In other words, each
discrete pixel is of one unique color (within the limitation of the
color dye clouds which are created by the photographic grain and dye
formation).

Inkjet prints are not continuous tone, in that each "pixel" is created
through a combination of discrete dots made up of the limited ink colors
the printer uses. Inkjet printers use the fact that our visual acuity
is limited and beyond a very minimal distance our eyes can no longer
resolve the individual dots. We therefore accept the addition of white
"space" around or intermingled with the dots (the paper in the case of
inkjet prints) as a lighter color. With some thing like the Lightjet,
each pixel is recorded as on of several thousand levels of brightness
and millions or even billions of color levels. The limitation in this
process is probably the ability of the paper dye layers to define the
values, rather than the ability of the lasers to produce the color
levels (I believe the Lightjet is capable of 36 bit depth).

Today's inkjet printers have the added feature of being able to supply
variable dot sizes. Epson does it by literally altering the vibration
frequency of the piezo, which changes the fractionation of the ink
droplet, others use several different size nozzle outlets. Besides the
several confounding factors that Frederick mentioned (dot deformation
due to the head velocity, number of ink colors necessary to create a
gradient, paper type creating differing dot gain, etc) the ink density
required for coverage determines what size dots are used, and in fact
they may vary within a small area, making the only measurement possible
empirical, and variable depending upon where the measurement is taken.

The good news is that the overall effect for us, using our human eyes,
is that today with the nearly one picolitre minimum volume dot, we can
no longer focus on the smallest discrete dots, unless one is as myopic
as I am ;-).

The other good news is that although the color gamut may be somewhat
compressed, at the size and frequency of dot modern inkjets can produce,
particularly with dye colorants which are transparent, a 4 color CMYK
printer can produce a very acceptable print, meaning if the
manufacturers wished to, they could make a quite inexpensive to operate
(and manufacture) printer, using only 4 heads and 4 colors.

Art

 

[Marc Wossner]

Thank you Art for the thorough description of what´s going on! But
thinking of all that, I come back to a point that still is not quite
clear to me.
As I mentioned in my first post I read that inkjet printers have
something like a native resolution (720ppi for Epson and 600ppi for
Canon). But how do I have to imagine the relationship between this
native resolution and the ppi value of the image being printed?
Is ist something like a virtual (read: not physically existing) grid
that the printer subdivides the area it has to print on in? This would
then be a kind of matrix which again contains cells that can be either
marked or unmarked with dots of ink to produce a halftone.

Marc

 

[frederick]

Besides the
several confounding factors that Frederick mentioned (dot deformation
due to the head velocity, number of ink colors necessary to create a
gradient, paper type creating differing dot gain, etc) the ink density
required for coverage determines what size dots are used, and in fact
they may vary within a small area, making the only measurement possible
empirical, and variable depending upon where the measurement is taken.

IIRC the dot deformation I saw wasn't in the direction that the head
moves. I don't know why. On photo paper, at ~ 100x, the dots appeared
well defined - not smeared.
Size of dots is an advantage that Epson have - with Piezo heads, they
can control dot size by frequency and amplitude of signal - so one
nozzle can apply many (9?) different sized droplets. (Piezo also fire
the nozzles at a much faster rate rate - especially continuously)
Canon work around the essentially binary nature of thermal heads by
having several different sized heating element per nozzle, HP by having
many more nozzles. Typically an HP printer has 10 times as many nozzles
as an equivalent epson to print at similar resolution at similar speed.
This is perhaps why, despite advanced technology built in - to monitor
and service nozzles - the new HP pigment printers seem to suffer no less
banding than competing and much simpler epson models.