Ken Norton wrote:
>> As the angle varies from straight ahead, the focal distance increases and
>> the image gets smaller. The effect is not particularly obvious with many
>> subjects. With a small, even, linear band like the sea in this image, it
>> becomes very obvious, appearing larger in the middle and tapering off in
>> both directions.
>>
>>
>
> I do not agree.
>
You are wrong, perhaps because you don't understand what I said. On the
theory that misunderstanding may be at least partly my fault, I elaborate.
First, I wrote "focal distance", i.e. distance from lens to subject, not
focal length, i.e. distance from lens to film plane. Perhaps I should
have said subject distance?
Consider a wide mural on a flat wall. With the lens nodal point at 10'
from the center of the mural, the distance from node to subject at 45
deg. to left or right is 14'. At 70 deg, the subject distance is almost 30'
> It is true that a single wide-angle lens that is corrected to provide a
> linear image increases the focal length the farther from the center of the
> image. This is very apparent with an ultra-wide lens.
>
Indeed - and that is the good news. As the subject distance increases
with angle, so does the distance from lens to film. In fact, per the
principle of proportional triangles, the effective magnification exactly
matches the increased subject distance. The effect of widening out of
subjects near the edge of very wide angle images is a different
geometric problem. Look at that head at the edge; it's too wide, but the
right height.
> However, in the case of a panorama camera with rotating lenses AND a curved
> film plane, there is no increase in focal length other than the minor change
> from top to bottom. But the laws of projection would nullify any
> focal-length change when the sensor or film plane rotates equally.
And that's the bad news. Since the lens is an ordinary design and only
the central part is used in the swing design, the focal distance doesn't
change. So the image gets progressively smaller as the angle from
straight ahead gets larger.. You can do the optical calculation - and
can see the effect here. <http://en.wikipedia.org/wiki/File:CH1.jpg>
Search for these images on the web, the effect is subtly to glaringly
obvious in many of them. Search for "cigar distortion" and you will find
descriptions, diagrams, etc.
> If the film plane is flat and the lens rotates, then focal-length increase
> occurs. When this happens, you also get extreme "ripping" distortion the
> closer to the edge and corners as is so common with extreme wide-angles.
>
Again, a different phenomenon, see above.
> When you do a panoramic photograph with a standard digital (or film) camera
> and you attempt to rotate the camera-and-lens around the rear nodal point,
> again there is no increase in focal length as the sensor/film is rotating
> equally with the lens. You will always get some distortion between
> overlapping frames, but this is the fault of the lens being used and is why
> it is best to use as close to a "normal" focal length as possible when
> taking the individual photographs used in a stitching.
>
Again, I differ. The effect, absent correction, as in Nathan's beach
panorama, is "cigar" distortion. The reason you see it so little in
digitally stitched images is that the software corrects for it, mostly
automatically. Perhaps a look at part of the PS dialog for merging
photos will illustrate.
<http://galleries.moosemystic.net/MooseFoto/index.php?gallery=Miscellaneous&image=PanoramaDistortion.JPG>
The default selection of Auto does a good job with many stitchings.
> A variation on this theme is when you use lens-shift instead of rotation for
> shooting the frames used in the panorama. As a lens is shifted to the side,
> the greater the effective focal-length.
>
> I maintain that the reason why you see the bulge in the middle of panoramas
> is that the subject is usually closest in the middle.
Exactly! Closest is largest. That's what I said, or meant to say and you
misunderstood.
> For example if you are standing on the curb of a street and have the camera
> pointed directly across the street, the distance to the other side may be
> only 25 feet, but you might be seeing a mile or more down the street each
> way. Of course the cars, trees and pedestrians are going to "bulge" at you.
>
> The only way to avoid the bulge (and linear distortion of straight lines) in
> these photographs is to assemble and display the panorama in a curved plane
> around the viewing position. This is NOT a capture issue, but a display
> issue.
>
Not if you use the digital option.
Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|
