ASTRO: Size of Files for Sharp Prints

Hello everyone: I have a question regarding file size and print quality. I
am currently preparing a publication involving wide-field astrophotographs.
I am working with tif files of 3200 x 3200, and about 60 megabytes. The
prints will be in a "coffee table" sized book , but let's just say around
11x14. Are digital files of the size mentioned above capable of producing
prints of high quality that are 11x14?

Thanks for your insights,

-Charles

"Charles Shahar" wrote in message
...
Hello everyone: I have a question regarding file size and print quality. I
am currently preparing a publication involving wide-field
astrophotographs.
I am working with tif files of 3200 x 3200, and about 60 megabytes. The
prints will be in a "coffee table" sized book , but let's just say around
11x14. Are digital files of the size mentioned above capable of producing
prints of high quality that are 11x14?

Thanks for your insights,

-Charles



Reproduced as 11" x 14", you can calculate that the resolution will be about
290 dpi x 230 dpi.

If you crop the square image it 11 x 14, it will be 230 dpi. If you limit
yourself to square images of (say) 10" x 10", the resolution will be 320
dpi.

These are the sort of resolutions you get with laser printers; glossy art
books need 1200+ dpi for high quality images.

They will probably look good but not great. I would try some to see - take
some images down to a print shop and ask them for the highest quality output
they can generate, and have a look.

"Peter Webb" wrote in message
u...

"Charles Shahar" wrote in message
...
Hello everyone: I have a question regarding file size and print
quality. I
am currently preparing a publication involving wide-field
astrophotographs.
I am working with tif files of 3200 x 3200, and about 60 megabytes.
The
prints will be in a "coffee table" sized book , but let's just say
around
11x14. Are digital files of the size mentioned above capable of
producing
prints of high quality that are 11x14?

Thanks for your insights,

-Charles



Reproduced as 11" x 14", you can calculate that the resolution will be
about
290 dpi x 230 dpi.

If you crop the square image it 11 x 14, it will be 230 dpi. If you
limit
yourself to square images of (say) 10" x 10", the resolution will be 320
dpi.

These are the sort of resolutions you get with laser printers; glossy
art
books need 1200+ dpi for high quality images.

They will probably look good but not great. I would try some to see -
take
some images down to a print shop and ask them for the highest quality
output
they can generate, and have a look.
Not quite true.
It is important to understand the difference between 'mono' dots, and dots
with a range of intensities. For instance, a typical phototypesetter used
for a newspaper, will have a resolution of 1200dpi. However each dot, is
only on or off, and is used by the machine to 'print' a reprodution of the
old litho masks. A pattern typically 8 or 10 dots 'square', is printed to
produce a number of different shapes and fill intensities. The paper is
normally printed on presses that have perhaps a line reproduction ability
of only around 100lpi (many papers use as low as 60lpi). High quality
'photo' prints still only use presses capable of about 200lpi, but the
phototypesetter, will then run at 1800 to 2400dpi, and produce a mask with
several thousand shapes available at this resolution. The setters also so
things that normal computer printers don't do, such as producing shapes
with one 'side' missing, when printing on the edge of a vertical line.
However incoming colour photo artwork, still only requires the line
resolution, not the dot resolution of the typesetter. This difference is
why (for instance), a dye sublimation printer, which can produce hundreds
of tones for each 'dot', yet only has a resolution of 200dpi, produces
pictures that are as good (or better), than laser or inkjet printers with
four times the resolution.
Now in printing, the pattern, usually follows a layout, more like the CCD
pattern in the Fuji cameras (with each alternate line displaced). This
improves the diagonal resolution of the page, so when reproducing images,
it is normal to use images that are at least 50% 'oversampled', relative
to the grid mask being used. Hence images at 300dpi, are normally
considered the acceptable minimum for 'high quality' printing, and 230dpi,
will match the work produced in most reasonable quality magazines. The
images being referred to, have far more colour depth than will be
reproducable. In most printing, you will be 'lucky' to get a range of
intensities in the order of only 8bit deep, with only about 1000 colours
distinguishable. From the sizes given, these are 16bit images for each
colour plane, with far more colours present. It will be well worth
spending the time to bring the colour range down to a printable level
yourself, rather than relying on the print shop to do it. However you then
also have to 'beware' that most normal monitors are not sufficiently
accurate to give a reasonable representation of the printed page (look at
colour calibrated monitors for this...).
The print sizes are 'borderline' for the resolution quoted. This also
depends on whether the image itself matches the resolution given. If the
actual picture is oversampled relative to the imaging system, then the
blurring this involves will get worse as the image is enlarged.
Most better print shops, will have proof printers, and having an image
reproduced on this will show what can be expected.

Best Wishes

Charles Shahar wrote:

Hello everyone: I have a question regarding file size and print quality. I
am currently preparing a publication involving wide-field astrophotographs.
I am working with tif files of 3200 x 3200, and about 60 megabytes. The
prints will be in a "coffee table" sized book , but let's just say around
11x14. Are digital files of the size mentioned above capable of producing
prints of high quality that are 11x14?

It depends on how they're being reproduced. For offset printing on
coated paper, with conventional halftone screening at 150-175 lpi,
your TIFFs should come off pretty well at sizes from about ten to
fourteen inches square. The usual rule of thumb calls for a pixel
count of 1.5 - 2 times the screen ruling, so 14" is about the largest
you'll want to go without upsampling. If "stochastic" screens will be
used, or some printing method other than offset, the requirements may
be different; it's always a good idea to talk to the printers.

Another important consideration is how to convert from RGB to CMYK,
whether by 'traditional' UCR/GCR or using a colour-management system;
if you're expected to provide CMYK files, particular attention should
be paid to the maximum ink coverage in the black background -- again,
ask the printers for their preferred specifications.

--
Odysseus

Peter Webb wrote:

Reproduced as 11" x 14", you can calculate that the resolution will be about
290 dpi x 230 dpi.

That would involve distorting the image to a completely different
aspect ratio, and would turn round stars into ellipses: not
recommended. Cropping is pretty much the only way to go unless
'artistic effect' is all that matters.

If you crop the square image it 11 x 14, it will be 230 dpi. If you limit
yourself to square images of (say) 10" x 10", the resolution will be 320
dpi.

These are the sort of resolutions you get with laser printers; glossy art
books need 1200+ dpi for high quality images.

The imagesetter producing the film or plates must have a very high
resolution (technically "addressibility", but that's another story)
but the dot patterns it produces are strictly black and white, one
bit deep. Here we're talking about a "continuous-tone" image that
will be rendered by a fine pattern of these dots, on a much larger
scale. Note that in the usage of the pre-press trade "dpi", dots per
inch, refers to the grid of laser 'aiming points' (the dots
themselves are typically 15-25 microns in diameter); "lpi", lines per
inch, is for the frequency of a halftone screen, in each of whose
"cells" a certain number of pixels are turned on to simulate a given
grey-level or colour density; and "ppi", pixels per inch, describes
the sampling frequency of a source image.

In order to produce 256 levels of tone (eight bits of depth) -- more
than most presses can reproduce -- with conventional halftone
screens, in theory there have to be at least 256 laser dots in each
halftone cell. In practice considerably more are needed for various
technical reasons I won't go into here (unless you're really
interested), but ignoring that, each halftone spot needs a
16x16-pixel grid to cover the range. So running 150-lpi screens
demands a resolution of 2400 dpi at the very least. Where I work we
produce our finer screens (150-175 lpi) with our imagestter set to
3556 dpi; OTOH we run "line art", solid black (or spot colour) and
white with no shading, at 1270 dpi. (These figures are rounder in
metric.) When we scan line art, we generally sample at 1200 ppi or
higher where the quality of the original will support it, because in
principle every pixel of a one-bit-deep source file can be rendered.

Anyway, a continous-tone photo is reproduced by simulating variation
in density with (solid) halftone spots of various sizes, and it's the
screen ruling, not the underlying laser-dot grid, that determines the
effective resolution of the printed image. When rendering an image
the RIP (raster image processor) averages the sample values covered
by each halftone cell to determine the size of spot that will appear
at that location, so no detail smaller than a certain size can be
preserved. As I mentioned in my earlier posting, the usually quoted
guideline calls for the sampling rate (ppi) to be 1.5X to 2X as great
as the screen ruling (lpi), giving about 2X to 4X oversampling.
There's no benefit to going higher; if you download a 1200-ppi image
to be rendered at 150 lpi, the size of each spot will be calculated
from the average grey value in a block of 64 pixels, and in fact it
will likely come out looking worse than a properly sharpened image of
appropriate resolution -- beside the waste of disk space and time
entailed in storing an needlessly enormous file and pushing it around
a network.

--
Odysseus