| [Thread Prev] | [Thread Next] | [Thread Index] | [Date Prev] | [Date Next] | [Date Index] |
Tom DeVrie wrote:
>One of the limitations for electronic journals in paleontology is the
>resolution of scanners for photographic plates. I have yet to see an
>electronic graphic with the clarity and resolution of a darkroom-produced,
>traditionally-published figure in JP.
This has nothing to do with digital versus non-digital techniques. The
reason JP has such outstanding illustrations is that Allen Press uses very
fine halftone screens (down to 300 lines per inch, I believe) and then
reenhances the lost contrast by printing a layer of varnish on top. This is
costly, but the result is superb. Quality costs (or have I already said
that?). And we could do the same with illustrations pulled through a
scanner, if we like. The use of digital techniques improves the printed
image, if anything, because you have a much better control of tonal range.
(Though calibration is crucial, and the final printing procedure has to be
up to snuff, or you can have bad results just as with the traditional
techniques.)
>How much resolution is necessary to produce a photo that is (a) adequate or
>(b) excellent for electronic dissemination? 300dpi? 600 dpi? 1200dpi? More?
Less. To compare with printing with a halftone screen, you only need to set
the scanner resolution to 1.5-2 times the resolution of the screen. In
other words, if the journal prints at 150 lines per inch (lpi), you set the
scanner at 225-300 dots per inch (dpi). Any higher resolution than that is
just waste of storage space and computer power - you won't see any
difference in print. With images destined for web-posting, you could even
do with less - a pixel resolution of 150 dpi will show you about as much as
one printed with a 150 lpi halftone screen. Of course, if you enlarge the
picture too much you will start seeing pixels, in the same way as you see
dots when you enlarge a print. But images posted in too high resolution are
often aggravating with today's rather overloaded networks. One common
solution is to post the picture in several resolutions: one for quick
viewing on the screen, provided with a link to others for more
time-consuming downloading.
Web-posted images should still be good enough to allow printing on paper
(which a lot of users will still want to do), so you should be sure to
include not only the low-resolution ones.
If you scan lineart (drawings, etc.) you actually need higher resolution
than for photos. For high-resolution printing, you normally shouldn't go
below 800 dpi, or the oblique lines may appear jagged. For screen viewing,
you will have to deal with the inherent jaggedness of the computer screen
anyway (most screens are less than 100 dpi), so this may not be a big
matter. Again, several resolutions can be included, and images that take
minutes to load won't be popular on the web.
Another matter is the scanner's pixel depth, i.e. the number of bits per
pixel. A electronic greyscale image usually uses 8 bits, i.e. 255 shades of
grey. Modern scanner do that without problem. However, it's better to use a
scanner with higher pixel depth, if possible, even if the final image is to
be saved as an 8-bit file, because that allows for better adjustment of the
tonal range. For lineart you need only 1 bit per pixel (it's either black
or white), which compensates for the higher resolution needed.
So modern flatbed scanners are quite sufficient for the purpose.
>At the other end, what kind of screen or printing resolution do we need for
>using the electronic photo? The 72dpi on my Mac screen? Do I need to invest
>in a Tectronix high-res monitor? What about the printer? 300dpi? 600? How
>many levels of gray? Color?
Obviously it would be nice to have a resolution approaching what we are
used to from print, but even the common 72 dpi Mac screen does OK - you
only have to enlarge the image a bit so as not to loose any pixels between
the screen dots. But if you have a small monitor, you may not then see all
of the picture at once.
The printer is the weakest link, because an ordinary laser printer prints
only black dots and uses a halftone screen pattern to build up greyscale
images: smaller spots for light and larger spots for dark. The problem is
that the spots have to be built up by aggregates of black dots at the
printer's own resolution, and in order to build up spots of 255 different
sizes (to represent the 255 shades of grey) you need a linear resolution of
the printer that is higher than the halftone screen frequency by about the
square root of 255, i.e. 16. In other words, a 300 dpi laser printer, in
order to print all 255 greys, would need to print with a halftone screen of
less than 20 lpi, which will look more like a polka-dot pattern than an
image. So grey levels and/or resolution have to be sacrificed.
To print a photograph with a 120 lpi screen and 255 tones of grey (which is
about the lowest acceptable) you need an imagesetter or a laser printer
that is capable of about 2000 dpi. A better solution then would be to use a
continuous-tone printer (e.g. one using dye-sublimation), which can print
actual greys (though text will be less crips with such printing). These
gadgets are expensive, though, and hardly anything you put on your desk.
Cheaper solutions are on their way, though.
>These technological considerations would affect my decision to submit and
>subscribe to an electronic journal.
We're still very much in the transitional phase - it's only recently that
it has even begun to be realistic to go for this kind of solution. For
shere reading convenience you still have to beat paper - and I expect paper
not to be replaced, but to be complemented by electronic dissemination.
Stefan Bengtson _/ _/ _/_/_/ _/ _/
Department of Palaeozoology _/_/ _/ _/ _/ _/_/ _/_/
Swedish Museum of Natural History _/ _/ _/ _/ _/ _/ _/ _/ _/
Box 50007 _/ _/ _/ _/_/_/ _/ _/ _/
S-104 05 Stockholm _/ _/_/ _/ _/ _/ _/
Sweden _/ _/ _/ _/ _/ _/
tel. +46-8 666 42 20
+46-18 54 99 06 (home)
fax +46-8 666 41 84
e-mail Stefan.Bengtson@nrm.se
Partial index: