Post
by Bill Jurens » Tue Jan 28, 2020 5:52 am
Attempts at any sort of photogrammetric analysis of this photography are, at best, extremely "iffy". I visited the Naval Historical Center website and noticed that their digitized images seem to contain what appears to be a slight curl on the right-hand side, which suggests that they were not made on a flat-bed scanner, but simply represent rather offhand photos of prints in the collection. This sort of thing is not to serious when the image is being used just for visual reference, but more-or-less destroys any ability to do photogrammetric measurements, where errors in hundredths of a millimeter can often be significant. Proper photogrammetry is REALLY picky work.
I doubt if there is anything further that can be done without access to the original photography, or at least to something close to that. Ideally, one would have a negative, although even then the quality of the measurements which could be extracted therefrom is problematical as very few lenses, even for high-quality cameras, are of photogrammetric quality. In most cases hand-held cameras are really designed to give one the maximum amount of light-gathering capability, which often comes at the cost of fairly significant geometric distortion, particularly close to the edges of the frame, and particularly when the lens is used nearly 'wide open', which, due to the low film speed available at the time was likely the case in Denmark Strait. Further, most of these lenses were not really well corrected for errors due to variations in subject coloring insofar insofar as almost all photography back then was done in black-and-white. In most cases this represents a good tradeoff because small geometric distortions are rarely noticed or considered important in 'snapshot' photography. There's another problem, too. Most photogrammetric cameras are of fixed focal length and permanently focused at infinity, because most photogrammetric subjects are nearly at infinity anyway. Adjusting the focus away from infinity, which most 'snapshot' photographers would consider essential, causes problems because the projected geometry changes as the lens elements move. Many years ago, I did a variety of tests with various consumer grade cameras to determine if they could be used, in a pinch, to substitute for a photogrammetric camera. OK for approximations, not for anything much better than this. Focal lengths were only approximate, within perhaps 1.5% of the 'published' value, so a nominally 150 mm lens might have an effective focal length of something between 147 and 153 mm. And people kept changing the focus. Photogrammetric cameras have both a published 'nominal' focal length, of (say) 150mm, but their actual focal length, measured carefully in the factory, and chosen to minimize distortions, is always something different, and is stamped on the lens, to a hundreth of a millimeter or better, e.g. 152.56mm
I suspect, but have no way of knowing for sure, that the original negatives -- which were of little use militarily --were retained by the photographer, who probably allowed a limited number of prints to be made in the ship's darkroom to be given out at souvenirs. Other images -- probably better ones -- were likely supplied to various and sundry news bureaus, who would have screened them for publication. So, what we really have are a set of surviving photographic prints, in various sizes and scales, with nothing to track back to the original image. If negatives DO exist, it is quite likely that they are not real negatives at all, but so-called 'copy negatives' made by re-photographing a print. In such cases, little or no effort was usually made to maintain the photogrammetric geometry of the subject. The original negative was likely cropped in printing if nothing else by the negative holder in the enlarger, and -- if copy negatives were made -- may have been cropped again slightly to fill the new negative with the old picture. What's even worse, in many cases old 'analog' negatives have been digitized, which can further degrade the image significantly. Copying a photo NEVER improves the image -- although the copy may look more pleasing to the eye, you can be sure it contains less information than the original did.
It would be useful if someone could visit the German archives to see exactly what, if anything survives. I suspect that at best they have copy negatives. But it would take an archivist, and perhaps a skilled photographer experienced in pre-digital imagery to tell for sure.
Analysis via receding lines converging on the principal point can be confusing, and is largely dependent upon the geometry. Lines are only radial when the subject is vertical and the photographic plane is perpendicular, or nearly so, to the camera's line-of-sight. On conventional aerial photography, one can estimate the position of the principal point by projecting things that are known to be sharp and vertical, e.g. the vertical corners of a building, towards the center. But this really only works when the corners are actually vertical. And it's only an approximation at best. Sometimes pretty good, sometimes pretty bad. And, it only tells you, at best, where the camera was pointed at the time of photography, which is not really enough to go on for further work that is not very highly speculative. Going farther than this, though it might be interesting in an academic sense, would take a great deal more text and explanation, basically moving us into the beginnings of a formal course in photogrammetry.
As I once taught this stuff, I can go on more-or-less forever...
Comments, as always, welcome.
Bill Jurens
