I'm still not sure, Carlos, that I'm following you. Presumably you took the
picture knowing you'd use a perspective crop on it, so there's no loss of
information, other than resolution, from what was intended.
Or alternately you mean that if you didn't have cropping in mind when you took
the picture, cropping it later will leave something out – as is always the
case. As Jeff said, you have to plan ahead and leave enough foreground (or
enough sky above your church cross) for the crop.
Anyhow here's an example. I tried half a dozen places in the alley before I
got the framing I wanted. The image was taken so that after perspective crop,
(a) the 'E' in 'Hope' would be included, along with its light standard; and (b)
the diagonal wire at the left would make contact with a bit of building and not
just wander off the left margin. The perspective crop has of course
interpolated pixels at the top, and in that sense there's technically a loss of
resolution, but with no visible effect. All intended information is included;
nothing left out or lost.
But back to the original post: the problem was probably just camera angle. As
in this photograph, you get more perspective convergence if you shoot from the
waist or chest. If you want a straightforward representation of architecture,
it's best to use a prism and shoot from eye level, which will end up looking
more natural to the viewer. (Or abandon Rollei for view camera or tilt-shift
DSLR lens, but that's a whole different story.)
It's interesting, though, that commercial photography and sometimes even
architectural magazines don't always correct for perspective. We have a kind
of secondary vision: the eye tricks us into seeing verticals when we walk
around in the world, but people have become so accustomed to viewing wide-angle
camera shots that they accept some convergence of verticals in photographs.
The 'tippiness' becomes part of the dynamics of the composition.
Kirk
PS: In practice I often take out some vertical convergence in an image with
Photoshop, but also leave some.
> Date: Mon, 12 Dec 2011 15:15:48 -0300
> Subject: [rollei_list] Re: Verticals at edge of frame
> From: cmfreaza@xxxxxxxxx
> To: rollei_list@xxxxxxxxxxxxx
>
> 2011/12/12 Kirk Thompson <thompsonkirk@xxxxxxxxxxx>:
> > "Loss of image information" should not be a significant problem. If you
> > make a 16-bit scan at 3200 ppi, you have enough info for a print of about
> > 24x24" @ 300 ppi. Hardly a problem in any reasonable perspective crop -
> > unless you're trying to turn a pyramid into a square?
>
> The image resolution has nothing to do with the problem, I meant you
> could lose and in fact you lose an area of the original image you are
> correcting, it could be significant or it could be no significant for
> the final result, it is more significant for close-up images and it
> is more significant if the perspective error is big. Another sample is
> to correct the perspective for a church with a cross above the tower,
> you could lose or you could crop the cross after to correct the
> perspective with PS or similar software, it depends about the
> correction amount needed and the image original composition, perhaps
> you lose nothing, only a bit of sky or perhaps you lose a significant
> detail; if you lost the cross after the correction, you can't recover
> it, it has nothing to do with the image resolution.
> I only pointed out the difference between the PS perspective
> modification tool and a view camera.
>
> Carlos
> ---
Attachment:
Anchor&HopeScan.jpg
Description: JPEG image
Attachment:
Anchor&Hope.jpg
Description: JPEG image
