Re: National Geographic vs. Canon print paper

sheepdog 2007 added these comments in the current discussion du
jour ...

You lost me here, I'm afraid. When I am scanning, I set the
DPI (correct usage of the term here) to produce the size
image I want in pixels).

Image size has two components, pixel dimensions (H x W) and
resolution (px/in). For now, let's leave compression out of
it, so this is an uncompressed TIFF file we're discussing. Our
example is a square image that's 200 pixels on a side. If you
change the res (using software) to 100 px/in the physical size
will be 2" x 2"; make the res 200 px/in and it's now one inch
by one inch. At 400-res the image is a one-half inch square.
You still have exactly the same amount of picture information
you started out with. Now let's talk scanning.

You always want to scan at the highest resolution you
anticipate using in the final image. If you were working on a
scan whose end use would be gravure printing in a car catalog,
you'd be scanning at 300-400 px/in or even higher. For viewing
on a monitor (desktop publishing, email attachments, web
pages) 72 px/in may be high enough, although your workflow
probably should allow for Photoshop or other processing and
that will probably work out best if you start with "extra"
picture information and throw some away as a part of saving
the final image.

I understand the basics of scanning pretty well. When scanning
half-tone vs. continous tone photos, I actually overscan from at
least 2x to 3x or even more than I need for the finished size. I
have found emphirically that the process of resampling down helps
to eliminate the "noise", actually geometricaly regular patterns,
before I need to get into noise reduction.

I scan Kodachromes at 300 res, clean them up, color correct,
tweak contrast, sharpen, etc., at that res and then do one
"Save As" for archiving of the layered, 300 px/in file,
followed by a "Save As" of a 72 px/in .jpg for web usage. In
my archives (CD or DVD) I have files whose names start with
the date and include a version number right before the file
extension. So "061203.tif" lets me know that's the unaltered
original that produced both "061203_04.psd" and
"061203_11.psd," which are layered Photoshop Documents which
have had different work done on them, and I may be able to see
the changes by referring to an iView catalog for the folder
containing them (which I know was created on Dec. 3, 2006) or
I may have to open them to see what I did.

Sorry, guess it is one of my dense days. I still don't understand
why you scan at only 300 DPI (if that is what you mean), unless
the negs/slides are the large cut-*** kind. For 35mm, most
people find that somewhere in the 2500-5000 DPI range better able
to extract all the image information available, then resampling
down to whatever they think is "right" for their purpose(s).

The main thing is, if you scan at 640 x 480 by 72 res that
will be a nice compact file for an email attachment,
especially when it gets compressed as a JPEG. but if you later
decide to blow it up to an 8 x 10 on your printer, you'll have
to rescan.

I'm further confused by your last statement. Are you saying
that you change the PPI/DPI to 72 for web and 144 for the
printer or is your dimension "300 px/in" different than what
I would call PPI (Pixels Per Inch).

I hope I explained targeting the different res for different
end use adequately. I write "px/in" for "pixels per inch" to
avoid the ambiguity of dpi vs ppi vs lpi, which is how this
whole thing started =')

Again, please excuse my denseness, as I must've missed your
intent. To me, being the simple engineer that I am <grin>, when
scanning, DPI is the correct term while when printing from an
existing image in pixels, then PPI is the right nomenclature.

As to LPI, I understand conceptually what this is about wrt the
printing biz but have never found any particular use for it when
either scanning photos or slides or in digital photography, so I
stick with the traditional DPI for scanning input and PPI for
printing output. But, as I suspected, everytime this subject
comes up, there is quite a long debate, with it's attendent
ambiguity as you observe.

--
HP, aka Jerry
.