See below.  Hope this helps.


----- Original Message -----
From: "Daniel Carrera" <[EMAIL PROTECTED]>
To: "gimp users" <[EMAIL PROTECTED]>
Sent: Thursday, March 27, 2003 9:16 PM
Subject: Re: [Gimp-user] [Q] Understanding the 72 DPI problem

> Thanks Kevin, I do feel much more informed now.  I still have a few
> questions, which I hope you could answer.

I'm not yet a complete GIMP expert by any means either.  There are MANY
features that I still know nothing about at all.  Glad to help out where I
can, and pay back a little of the help that other GIMP users have provided
to me.

> On Thu, Mar 27, 2003 at 09:37:27AM -0600, Kevin Myers wrote:
> > There are some VERY misleading statements on the web page referenced
> > regarding the specific question originally posted by the potential GIMP
> > in this thread.
> I looked at the site.  The content of the manual is great, but it deals
> with Gimp 1.2.2.  So I guess this is just outdated informatin.
> BTW, I am more than a "potential" user, I am a fully-converted GIMP lover
> :).  However, I've only ever used GIMP for the web, and I want to
> understand the issues involved in producing something for the printed
> press.
> > Finally, you can specify both resolution and size in the standard
> > dialogue, as well as in the preferences options for new files.
> I just noticed that, I can't believe I didn't see it before.
> My first question:
> If Gimp sees images as a matrix of pixels, what does it /mean/ for an
> image for have 100ppi?  Are you changing the ppi by just zooming in and
> out?
No, you're not changing the dpi of the actual image by simply zooming in and

> The concept of dpi/ppi only seems to come in when you print an image, so
> I'm confused.

You're right that printing is one important place where dpi comes into
consideration.  The other place is when you DON'T have the Dot for Dot
option selected, and want your image displayed on the screen at actual (i.e.
printed) size.  The GIMP allows you to provide screen size and resolution
information when it is unable to determine it automatically, so that it can
scale images appropriately for display.

> > I'd also like to bring up one very important GIMP advantage that I
> > see mentioned at all in the Photoshop vs. GIMP comparison.  The GIMP can
> > many cases handle LARGER images than Photoshop.
> This is very interestig.
> Elsewhere on the site give, it says that Photoshop is faster for large
> images and GIMP is faster for smaller images.  Is that still true?

Sorry, I can't answer that one.  Maybe some other folks can chip in.  All I
know is that for most of what I normally use it for, the GIMP is fast enough
for my needs, even with my huge images.  However, there are a few of the
filters that I use which are certainly slower than I would like (notably
unsharp mask).  But scaling, resolution, and color depth changes, and many
other filters are reasonably quick.  FYI, I primarily work with grayscale
and line art images (although the GIMP presently converts those to gray
scale too, which is a bit of a pain for me, since I then have to use some
other application to convert them back).

> > Photoshop (ALL versions) is limited to 32K pixels in any single
> > dimension, whereas the GIMP is only limited by available memory
> > (potentially including virtual memory).  I routinely process images
> > with over .5G total pixels, and over 300K pixels on one axis using the
> > GIMP.  Photoshop cannot handle these images at all.  In fact, I have
> > found very few other applications that can.
> That sounds huge.  Where do you use those images? posters?

Raster images of well logs for oil and gas wells.  These are essentially
strip chart recordings of various geophysical properties up and down a
borehole.  Even when the depth axis is scaled down significantly (e.g. 1
inch = 100 ft), these logs can still be tens or even hundreds of feet in
length.  In many respects as far as the GIMP as concerned, you could think
of them as very long, narrow posters.  I also deal with other very large
technical documents, such as maps of subsurface formation topography, and
geologic cross-sections.  These documents are typically palletized color,
300 dpi, and roughly 3 to 5 ft high by 4 to 8 ft long.

> How large should an image be so that it can be printed as a poster and
> still look good? (say, a 22x34in poster).

Not sure what exactly you're asking here.  But for true posters, with large
text that would be viewed mainly from a distance, I would think that 100 dpi
or even 72 dpi would probably be adequate.  On the other hand, I deal
primarily with images that are going to be viewed for up close detail, even
though the images are physically very large, unlike typical poster
applications.  So, my file size and resolution requirements are much larger
than normal.

> How much memory do you need to handle an image that large?
> Can a regular PC (e.g. Athlon XP 1800+) manipulate those images and be
> fast enough to be useful?

I am using a 2+GHz P4 machine with 1.5GB of RAM.  I have also used a 1GHz
Duron with 640MB, but that can really slow to a crawl on my largest images
that force swapping virtual memory to disk in order to handle the image
size.  Otherwise, the speed really isn't all that bad on the Duron either.
For reasonably normal size posters, I would guess that your Athlon XP based
system should be just fine as long as you have a decent amount of RAM
installed.  The amount that you really need depends on the exact size,
resolution, and color depth of your images.  If you can live with 256
palettized colors, then a 3 ft by 4 ft poster at 300 dpi (probably far more
than adequate resolution) would require 3*12*300*4*12*300 = 155MB, plus some
additional for the program itself and the OS.  I'd probably recommend at
least 256MB of RAM as a decent starting point for large, high resolution
poster work.

> Thanks for your help.
> --
> Daniel Carrera
> Graduate Teaching Assistant.  Math Dept.
> University of Maryland.  (301) 405-5137
> _______________________________________________
> Gimp-user mailing list

Gimp-user mailing list

Reply via email to