Re: digital question #3

From: Etienne Garbaux ^lt;>
Date: 08/01/04-01:35:33 PM Z
Message-id: <p05210600bd32ed54590e@[]>

Judy wrote:

> Etienne, would you describe "decently high quality" somewhat more
> concretely? I admit I can't help wondering about it in a day when Diana
> camera, pinhole camera, etc. are in vogue... (Or is that so last year???)
> Assuming the image is still readable, what are the benefits? And what are
> the pitfalls of interpolation?????

By "decently high quality" I mean that the print (on best-quality glossy
photo paper) is not obviously digital and compares well to a 35mm C-print
the same size. Many people think printing at 200 PPI gives this. I find
that it takes at least 300 to satisfy me. Of course, as you point out,
many people aren't TRYING to make maximum-detail (or maximum-resolution)
photographs -- where in my shot of the Brooklyn Bridge part of the
statement is being able to count the rivets, another artist may find that
such a degree of sharpness distracts from his or her statement about the
overall shape of the bridge. Still, maximum sharpness is a good criteria
for evaluation of (some of) the limits of our media, which exercise I
believe is worthwhile even if one's own work does not press that limit.

What are the benefits? One cannot create [actual] image information once
it has been lost, so it makes the most sense to me (in most situations) to
record lots of it and remove what is not important at a later stage of the
process. Of course, sometimes a particular loss of information -- like
loss of sharpness due to limited depth of field and the "bokeh" of a
favorite lens -- cannot satisfactorily be duplicated later, and must be
done at the time of recording.

The general rule is to keep everything at maximum resolution until the very
last step -- huge 48-bit files in my case -- and do any sizing as the last
step to match the image to the intended output medium. Reducing pixel
count is generally painless (that is, a 640x480 pixel image for the web
looks at least as good as a decimation from a 2000x3000 [6 Mp] image as it
would from a 0.3 Mp original [say, a cropped 1/3 frame from a 1 Mp camera].
On the other hand, if you want to print a 20x24" poster from that 6 Mp
image, you will either have to settle for a 100 PPI print or interpolate
pixels. If you want to print at 360 PPI, you'll need a 7200x8640 image (62
Mp). Photoshop's internal bicubic interpolation is not the best solution
here. There are several software packages and third-party Photoshop
plug-ins that claim to do magic in this regard, most of which (IMO) are
very little better than the native Photoshop bicubic interpolation.
However, I've seen one, in particular, that did a very creditable job
(though not, of course, the equal of an image from a now-hypothetical 62 Mp
sensor). (Sorry, I've forgotten the name of the software.)

Best regards,

Received on Sun Aug 1 13:44:33 2004

This archive was generated by hypermail 2.1.8 : 09/14/04-09:17:56 AM Z CST