*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
If I generate say a 800 x 800 pixel image in a program, it is just that.
It has no size in the file. It has no business being assigned a size in
any image header.
This is true for freeware rendering programs in general, which we use,
but the vast bulk of the 2D image "market" is digital photography, and
they tend to associate print sizes with images in a much more natural
way. One of the few areas that really doesn't care about print size is
a web browser, which only seems to take notice of the number of pixels.
Assuming you do minimal manipulation of your image before you are ready
to send it out the door, print size isn't really that relevant as long
as it is large enough, but:
If you are accustomed to using Photoshop to add your image label text,
then you'll find that Photoshop cares a great deal what your print size
is because it measures font size in points, not pixels. You can see
what I mean if you open Photoshop and create two new documents of
300x300 pixels, one at 72dpi and one at 300dpi and type the same text
with the same font into the two images. Photoshop renders the text at
different sizes, because it takes notice of the final print size of the
image, despite the number of pixels being the same.
I don't use Gimp but since it's supposed to emulate Photoshop in at
least some aspects I wouldn't be surprised if it did the same thing. I
do *not* want to initiate a Gimp vs Photoshop thread, however.
In general if you want the best results, it is best to feed the person
in charge of the printing something that is as close to their specs as
possible. Most graphics pros use Photoshop or Illustrator, or something
that does a pretty good impersonation of Photoshop, which writes image
size in the header where possible and has few qualms about using it.
Typically this involves sending them TIFF format although some of them
seem to prefer JPG. Typically this means 8-bits per color per pixel.
Even if TIFF were an evil format, it's an evil format that most graphics
pros are well-equipped to handle.
If you feed them the right size image (at least 300dpi for whatever
image size) and the right physical image size (or at least aspect ratio)
in the header, you tend to avoid having people with limited time on
their hands do things to your image that don't make it look quite so
pretty. Or, you can trust them completely in which case at least make
sure the aspect ratio of the crop is right.
Veering a little off-topic:
Of course at least some of this care and feeding aspect is meaningless
if one hasn't color-profiled one's monitor. Monitors vary a lot in
their representation of colors, and in my experience LCD ones are
perhaps the worst. If your monitor is a long way off standard then your
picture will inevitably look rather different in print than it will on
the screen, even taking into account the different properties of
monitors and print media.
In mentioning color profiling I have potentially opened a can of worms,
and before I re-seal it I want to mention that the same point applies
even more so to hard copies on color printers, if you use that to assess
the final image. Those things are so variable it's largely a joke.
Their major benefit is that they reveal when you've gone way out of
gamut for print media in your quest for a really nice image on the
computer screen.
Gamut is an even bigger can of worms than color profiling. I apologize
in advance.
Phil Jeffrey
Princeton
