Archived posting to the Leica Users Group, 1999/10/06

[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]

Subject: [Leica] Re: Leica Users digest V12 #44
From: "Robert Jones" <drrjones@flashcom.net>
Date: Wed, 06 Oct 1999 11:28:56 -0700

- ----------
>From: owner-leica-users@mejac.palo-alto.ca.us (Leica Users digest)
>To: leica-users-digest@mejac.palo-alto.ca.us
>Subject: Leica Users digest V12 #44
>Date: Wed, Oct 6, 1999, 9:01 AM
>

> Date: Wed, 6 Oct 1999 12:10:38 +0200
> From: "Anthony Atkielski" <anthony@atkielski.com>
> Subject: [Leica] Home photo printers and the digital darkroom
>
> From: Robert Jones <drrjones@flashcom.net>
> Sent: Tuesday, October 05, 1999 22:48
> Subject: [Leica] Re: Leica Users digest V12 #41
> >
> That's because of the difference between "machine dots" and "halftone dots."
> Halftone dots are dots of color of variable size that provide the different
> shades of color in the photograph.  Machine dots are smaller dots of fixed
size
> that are combined to produce the halftone dots.  The smaller the machine dots
> are, the smaller the halftone dots can be, but machine dots are _not_ the same
> as halftone dots, and the DPI figure quoted for a printer is typically the
> machine-dot size, not the halftone-dot size.
>
> For example, if you want 256 levels of gray in a photograph, you need halftone
> dots that can be of any one of 256 different sizes.  For that, you need at
least
> 256 machine dots per halftone dot.  Now, since halftone dots are two
dimension,
> you'll need an area (a box, say) equal in linear dimensions to the square root
> of the number of dots you require.  In other words, to print 256 different
sizes
> of halftone dots, you need a box that is 16 machine dots on a side (16x16=256
> dots).
>
> Given the above, the real resolution of a printer is equal to the number of
> halftone dots per inch (the "screen frequency"), not the number of machine
dots
> per inch.  For halftone dots of 16 machine dots on a side, with 1440 dpi of
> machine-dot resolution, the actual printing resolution is 90 halftone dots per
> inch or a screen frequency of 90 lines per inch (to put it into conventional
> printers' terms).
>
> This being so, you don't need more than about 150 pixels per inch in your
photos
> to use the maximum resolution of your printer, since the typical resolution
used
> for printing is equal to the screen frequency of the printer plus 50% or so.

Anthony,

Thanks for your comments. While I certainly am no expert, I think there is a
difference between high resolution ink jet output and halftone prints.

Your description of halftone dots and resolution are correct for halftone
(offset) printing, particularly for B&W prints in newspapers and books.
Halftone printing is a method used for low resolution output to get around
variable dot size and the lack of multiple shades of gray. Ink jet printers,
however, do not necessarily have to print in halftone formats. These
printers can vary the shades of gray as well as dot size for prints much
better than those obtained using a halftone screen. That is, the dot size is
only one variable when using MIS Quadtone inks, the other variable being
different shades of gray. This means that the 150 pixels per inch + 50% does
not apply (unless of course your output is to a halftone printer).

I have done tests with the Epson 1200 at various settings from 180, 240,
288, 360, and 480 (all integrals of 1440). There is a definite improvement
in clarity going up to about 360. I can barely see an improvement at 480 in
some fine details when printing 12x17 inch prints on 13x19 paper.

As for the actual resolution needed during a scan, it is proportional to the
degree of enlargement needed and what you consider the minimum amount to
send to the printer. That is, in my hands (or rather with my eyes), I want
to print at my optimum setting at 360 dpi and I want a final print that is
10X larger than my starting image (say a 1 inch portion of a 35 mm frame),
then I need to have an image scanned at 3600 dpi. If I scan at anything less
(as per this example) then there will be a noticeable degradation.
Similarly, one cannot re-interpolate with software (such as the re-sizing
function in PhotoShop) as this simply introduces artifacts (sort of like a
lousy zoom).

Certainly, there is a limit to what resolution is needed. It appears that
somewhere between 4000 dpi and 5000 dpi one is at the level of the grain in
the film, so going beyond that point makes little sense. (And the files
become too large; 4000 dpi files from 35mm are around 50MB per image.) This
is why, at 4000 dpi, 35mm film cannot be printed much larger than about
16x20 (the short side of the 35mm frame is about 1 inch or 4000 dpi; 4000
dpi/288 = 14X or and image that's 14x20 in size). This limit is basically
true for optical enlargements (at least I think most of us would agree that
enlarging a 35mm frame beyond 16x20 gives only acceptable, not outstanding
results). So it is curious, to paraphrase Michael Reichmann (who has an
outstanding web site on these issues at http://www.luminous-landscape.com/),
that both digital and optical enlargements come up with the same
limitations.

Regards,

Robert