[filmscanners] RE: dpi - formerly PS sharpening

["Laurie Solomon" <laurie@advancenet.net>]

Bob,

Since this has gone beyond the pale of what might be relevamnt to the list
and could result in another terminological free for all with everyone and
their brother jumping in with their two cents, I have decide to post this to
you privately in hopes that by keeping it between us it will remain civil
and not take on a protracted life of its own.  I did, however, feel an
obligation to be courteous and respond to your post even though you hinted
that you were finished with the topic.  In the end, we will just agree to
disagree; but we will know each others position and reasons for it.

>but I still maintain that there is a real difference between the
>'things' that define the image size - call them dots, pixels or samples -
>and the drops of ink that an Epson printer squirts onto the paper which
have
>nothing to do with image size.

You not only have a right to maintain that, but, of course, you are right
they are two differnt things.  One is a measure and the other is something
tangible that is measurable.  You are also entirely correct that that the
number of squirts onto the paper have no necessary relationship to the image
size in dimensional terms, although it may be related practically in an
empirical way such that to produce a satisfactorily rendered and readable
image you may need more squirts on the paper for a larger image size print
than for a smaller one.  However, you can have as many or as few squirts on
the paper as you like for any given image size dimension; but a 1 squirt 8 x
10 image may not look as good as a 720 squirt 8x10 image  or as good as a 15
squirt .8x1 inch image.

>If 'dpi' is used for everything, then when I send an image with 256 dpi to
>my printer, the printer converts it to 720 dpi in the driver and then
prints
>it at 2880 dpi in one direction and 1440 dpi in the other (using a most
>recent printer as the example). Does this make sense to you?

Yes, under a whole set of different scenarios. In such a case, the printer
via its driver is upsampling or interpolating from 256 dpi to 720 dpi for
any one of a number of reasons (e.g., device requirements, halftone screen
or dithering requirements, or halftone screen or dithering algorithm
application results, etc.).  Under your scenario, why would the printer
driver want to convert or translate 720 dpi into 2880 dpi in one direction
and 1440 dpi in the other rather than trnsforming the original 256 directly
into 2880 dpi in one direction and 1440 dpi in the other without passing go
(i.e, the 720 dpi conversion in the driver?  It could be that the 720 dpi
transformation to 2880 dpi in one direction and 1440 dpi in the other makes
no sense at all since the original 720 does not specify horizontal x
vertical specifications leaving one to assume that the per inch refers to
only one dimension and does not refer to square inches.  However, in
contrast to saying that it is sensless or a wrong use of measures,I would
suggest that there may be a number of reasons for this as well.

It could be that the 2880 dpi in one direction and 1440 dpi in the other
actually represents not dots per inch but ink squirts per inch independent
of the size of the squirt or the amount of ink it contains or the mark on
the paper that it leaves. Since dithering is involved each dot in the 720
dots per inch would require .5 squirts per inch across the horizontal; and
because of the step motor tractor feed of the paper vertically each
horizontal line of squirts would be a doubled resulting in 2880 squirts per
inch being laid down vertically.

It might also be that the printer specifications of 2880 dpi in one
direction and 1440 dpi in the other should not be in dpi at all technically
but that the company and its technical writters decided to use the
convention of dpi as the name for its unit of measure so as to give everyone
a known concept to relate to even though the whole notion of dpi may pertain
to uniform undithered dots in a halftone line screen as traditionally used
by press printers and not to the products of variable dithered inkjet
products.

>Is it not more meaningful to say instead, that I sent an image with 256
>pixels per inch to the printer which upsampled the number of pixels to 720
>pixels per inch (its native image resolution) in the driver, and then
prints
>it using up to 2880 ink dots per inch horizontally and 1440 dots per inch
>vertically (its real ink-drop max resolution)?

It is meaningful to me (although I am not sure I would go as far as to say
it is more meaningful).  It does sugggest something similar to what I
suggested above in responding to my own restatemetn of the question - e.g.,
"Under your scenario, why would the printer driver want to convert or
translate 720 dpi into 2880 dpi in one direction and 1440 dpi in the other
rather than trnsforming the original 256 directly into 2880 dpi in one
direction and 1440 dpi in the other without passing go...?"  But like my
suggestions, it does make an underlying basic presuposition.  Namely,
withiout using the step motor notion to account for it, it assumes native
pixels are rectangular in shape and represent lineally delineated
dimensional area as measured in terms of square inches/cms as contrasted say
to a circular shaped lineally delininated dimensional area.  This would
account for the translation of 156 or 720 dpi into a combination of
horizontal by vertical dimensions. But are they rectiliniar and do they have
dimension?  All that I have heard and read suggest that native pixels are
like points; they have no inherent dimensionality or shape.  That they
happen to display on a monitor as square or rectangular is just a
happenstance or artifact of the monitor display and not an actuality
inherent to them as phenomena.

>An image 'unit' - call it what you will - is NOT the same as a printer
>ink-drop, and the resolution of each is a completely different parameter.
>One resolution is the number of units per inch making up my image in PS and
>the resolution of the upsampled image in the printer, the other resolution
>is the maximum number of ink-drops that can be accurately squirted onto
>paper by the printer to represent those image units. The latter resolution
>is even different in two dimensions, whereas the former is the same.

In one sense this is a truism.  A unit of measure is not the phenomenon that
is being measures.  The problem is that the "image unit" and the printer
"ink-drop" a set forth as prime or primative concepts.  They have no finite
definitive specification as to their nature, scope, range, size, or weight.
They are like the concept of "infinity"  Neither are very precise in
quantifiable terms or in literal terms as to their defining boundaries such
that one can tell when one "image unit" or one "ink-drop" begins and another
ends except indirectly as an artifact of the measuring device be it monitor
display or hardcover printed page.  Moreover even the indirect presentation
of these units is not standardized.  How many ounces or ccs are in a
ink-drop; how wide are its boundaries?  The same can be said for the "image
unit" call it pixel or dot. It is the question of how many angels will fit
on the head of a pin.  The answer would consist of "it depends" (on the size
of the pinhead and the size of the angels among other things such as what
are the angels composed of solid or gasious materials).

In another sense this is an oversimplification which is not really as
clear-cut as you make it appear.  First, once again, Epson and other printer
manufacturers use the convention dpi when referring to the print quality in
their driver software; this is a single dimension dpi figure which the user
can often select to use one from among many but not always.  It is not a
vertical by horizontal dpi figure. it is not ppi but dpi which the print
driver targets as its intended print resolution for the image that was input
to the driver from wherever.

You are right; this is completely separate, although frequently connected,
and different from the number of ink droplets that are squirted in
accordange with the targeted resolution and the represented image; but the
latter per se does not represent resolution either that the printer is
capable of or that will obtain in the printed image produced.  The simple
number of squirts or drops or ink in any direction or number of directions
is not resolution.  The size in terms of uniformity or variablity as well as
lineal dimensionality are often the determinent factors of output
resolutions or printer resolution capabilities as well as the dithering or
halftoning algorithms, ink spread or gamma of the ink on the media, and how
many of those ink drops are identical duplicates except for the color of the
ink being put down such that 400 drops of ink might not actually be 400 dpi
but 100 dpi if 4 separate colors are laid down over the same inentical area
and in identical locations so as to produce not reolution but color
combinations.

I will not go into questions of whether or not the statement "One resolution
is the number of units per inch making up my image in PS...." is accurate or
just unclearly stated. Since you can only view the image file opened in PS
or any other program via an output device (typically a monitor display), you
can only speak to the resolution as it appears in the monitor display.  Here
many monitors and graphics cards also use the convention of referring to
resoutions as dpi not ppi; but aside from that and more importantly, what
resolution does your image have if it is displayed on a monitor having a
higher or lower resolution that that which Photoshop says your image is?  I
suggest that the output resolution would be the operative resolution in this
case and that the specified resolution in Photoshop would be irreleveant and
meaningless.  Similarly when one sends the image file to a differetn output
device for rendering or reproducing, the output devices resolution would be
the operative one and the PS specification would be meaningless and
irrelevant except as possibly a benchmark of some kind.

I will leave it here and not get into a discussion over the philosohy,
psychology and sociology of language and language use.

-----Original Message-----
From: filmscanners_owner@halftone.co.uk
[mailto:filmscanners_owner@halftone.co.uk]On Behalf Of bob frost
Sent: Saturday, August 17, 2002 12:09 PM
To: laurie@advancenet.net
Subject: [filmscanners] Re: dpi - formerly PS sharpening


Laurie,

I don't want to get into an endless argument over niceties of linguistics
either, but I still maintain that there is a real difference between the
'things' that define the image size - call them dots, pixels or samples -
and the drops of ink that an Epson printer squirts onto the paper which have
nothing to do with image size.

If 'dpi' is used for everything, then when I send an image with 256 dpi to
my printer, the printer converts it to 720 dpi in the driver and then prints
it at 2880 dpi in one direction and 1440 dpi in the other (using a most
recent printer as the example). Does this make sense to you?

Is it not more meaningful to say instead, that I sent an image with 256
pixels per inch to the printer which upsampled the number of pixels to 720
pixels per inch (its native image resolution) in the driver, and then prints
it using up to 2880 ink dots per inch horizontally and 1440 dots per inch
vertically (its real ink-drop max resolution)?

An image 'unit' - call it what you will - is NOT the same as a printer
ink-drop, and the resolution of each is a completely different parameter.
One resolution is the number of units per inch making up my image in PS and
the resolution of the upsampled image in the printer, the other resolution
is the maximum number of ink-drops that can be accurately squirted onto
paper by the printer to represent those image units. The latter resolution
is even different in two dimensions, whereas the former is the same.

Epson prints are made up of millions of ink-drops that look like DOTS. The
only DOTS in my images are the dots of red, blue and green phosphors on my
monitor screen, and their resolution is completely different again to that
of my image!

To call them all dpi seems to me silly when we have words that already are
in common usage and which describe them properly so that there is no
misunderstanding what one is talking about. Just because some people who
don't know better, and some that should know better, misuse technical terms,
does not mean that we should all fall in and misuse them. Education is
partly about learning the correct terms for things that need to be
distinguished.

I will say no more, but will continue to try to educate others to use
terminology that distinguishes between the resolution of images, monitors,
and printers.

Bob Frost.

----- Original Message -----
From: "Laurie Solomon" <laurie@advancenet.net>

>If you are going to use dpi for ppi, what are you going to call the real
>printer dpi? It's got to be called something different because it IS
>something different.

I am afraid I cannot take the above statemetns seriously.  There is no
"REAL" printer dpi and the difference is only a trivial difference just as
the difference between Kelvin, Celcius, anf Fehrenheit is or the difference
between meters and yards is.  It reminds me of the argument of an old lady
from Indiana who wrote the newspaper a letter to the editor against daylight
savings time arguing that we should not mess around with God's time.  Please
package up and send me 1 pixel, 1 sample, and 1 dot along with a point; and
while your at it, send along the precise dimensions of that unit of measure.
How large is a point, a dot, a pixel, a sample in order to be a point, a
dot, a pixel, a sample, and not something else?

Finally, I am responding to you in the spirit of friendship and not to put
you down.  I understand what you are saying and often follow a similar
course of action in my remarks to posts on lists; but when I do it is in an
effort to further the discussion of the substantive issues and not to
resolve terminological disputes or divert the discussion to a linguistic one
in itsown right.  In my response to Austin, I was not trying to take him to
task because I disagreed with him but because I felt the converstaion would
be diverted to a superficial issue.


----------------------------------------------------------------------------
------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title
or body

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body