Re: [SI] Shoot-In - More come trickling in "William Graham" wrote (to Ken):
"Michael Benveniste" responded:
"William Graham" replied:
Several points here, I think...
[1] William is right, the verticals do look OK. A vertical edge lined up
against the antenna of the foreground boat matches up just fine. So too
against the piling to the left of the flag in the background, and almost so
against the telephone pole beyond to the right of the flag. Topography is
rarely perfectly horizontal, but gravity is always perfectly (for
photographic purposes) vertical.
[2] Regarding "manipulations comparable to what those done in traditional
photographic processes" I think you would also be in the clear, as a
post-processing digital tilt correction would simply be simulating the
effect of rotating your darkroom enlarging easel slightly to square up a
composition prior to print exposure. With sufficient enlarging lens
depth-of-field, this technique also works in the z-axis to help correct
non-parallel verticals or horizontals resulting from incorrect film plane
orientation. (I know Mr. Benveniste knows this, but others may not have
considered it.)
[3] As for my case, I have tried the tilting feature in The Gimp 2.0 and
found it seriously wanting for my purposes. Perhaps PS is an improvement,
but I can't see how it could be. The problem is that my error is usually in
the 1-3 degree range of tilt needing to be corrected. Rotating for this
small of an amount introduces distortion which takes the form of an
introduced mushiness in the digitized reproduction.
I attribute this to the fact that the closer to zero degrees of rotation one
goes, the less scanned pixels the rotation algorithm has to work with. At
some point the logic would have to start skipping some pixels while shifting
adjacent ones (the stair-step effect), thus resulting in an overall loss of
visual sharpness. This effect would be analogous to the way pixel-rendered
lines look better at a 45 degrees tilt than at a 1 degree tilt.
Since my use of digital technology is self-limited to only reproducing
silver prints for sharing online, I limit myself to only attempting to copy
them as faithfully as possible - a seemingly impossible task at times.
Correcting tilts at this stage digitally would be akin to altering the
representation of the original work, thereby violating the Prime Directive
(sez Mr. Spock... ;).
Although I could back up one level and rotate the print itself on the
scanning bed, a better solution for me would be to back up two levels and
rotate the easel in the darkroom, thus creating a correct original silver
print from the get-go. In fact, I have done this on occasion when
necessary.
Even better would be for me to have learned of my tendency for crooked
horizons from generous comments in the SI, then subsequently paid attention
when my eye was at the viewfinder before I released the shutter, thereby
getting it correct on the negative itself. And maybe just quit
procrastinating and purchase that darned Type 'E' Nikon screen.
Ken
Ken Nadvornick
|
|
