Underwater wide converters - Discussion
The digital compact camera is remarkable for its ability to facilitate, and to some extent automate, the process of producing good-quality underwater photographs in a way which no film camera ever could. Thus, when technically mature models began to appear (in 2004), they immediately swept away the dreary focus bracketing, exposure bracketing, 3 good shots per film if you were lucky, tradition of Nikonos photography. Such cameras nevertheless go against a well established principle, which is that results will suffer if you try to do everything with a single (zoom) camera lens.
The first good compact cameras had a minimum focal length setting equivalent to 35mm. In other words, they couldn't do wide-angle photography at all. Put one into an underwater housing with a flat port, and the built-in lens will give a reasonable performance at any zoom setting. If you want wide angle however (and since water is cloudy, you do) you have to fit an adapter. The apparent logic of this situation is that the camera and housing combination is optically nearly-perfect, but lacking in FOV. Thus it is possible to fall into the trap of thinking that the adapter might possibly be eliminated, if only the camera could be given a wider minimum zoom setting.
The reality is radically different. It has been established over about 60 years of collective experience with 35mm film cameras (and is confirmed by theory), that there is a break point at about f = 35mm. At longer focal length, the use of a flat port is acceptable, because the optical defects introduced are not serious, and the reduction in FOV corresponds to an increase in magnification, which is useful for macro photography. At f < 35mm however, the pincushion distortion and aberration caused by a flat water-air boundary start to become offensive, and it is necessary to use either a dome port, or something akin to a Rebikoff corrector.
The unavoidable conclusion is that camera designers have rather spoilt things for the fixed-port underwater housing designers (even sometimes within the same company) by making zooms go to ever shorter focal lengths. Optical quality suffers at f < 35mm, it's not the housing maker's fault, and there is an issue of how to solve this problem in a way which users will understand and be prepared to accept.
Ease of operation is a primary consideration. Users want to use the maximum-wide zoom setting as a point of reference, and then fit an adapter to obtain even more coverage. Supplying an adapter which vignettes at the wide setting and telling people that they have to zoom-in to use it then appears to defeat the object. It does no such thing of course, because, as explained in the introduction article, the function of the adapter is primarily to correct for the water-air boundary, and secondarily to increase the coverage. This point however, sounds like an unscrupulous marketing ploy, and customers are likely to remain unconvinced. Adapter lens designers have therefore been forced to respond by trying to solve the 28 or even 24mm lens vignetting problem.
The results given in the preceding articles show that, unless the converter lens has an extremely large exit aperture, the best optical quality is always associated with the cameras (or camera lenses) which have a small zoom range and a moderate FOV at the wide setting. The reasons are that a large zoom range prevents the housing designer from placing the lens port close to the entrance pupil at wide-angle settings of the zoom, and a large FOV makes the adapter difficult to correct for aberration.
The results obtained are however, a little more equivocal than it might seem at first pass. It is perfectly possible that, with a well matched camera and adapter combination, the problem of having a large FOV at the camera might be surmounted. The drawback however, is that the camera manufacturers change their designs every few months, and so the details of the optical path are effectively undefinable. Thus, if we were to test every possible combination, we might find some good ones involving cameras having minimum equivalent focal lengths of 28mm or even 24mm. So far however, when using a compact camera in a housing with a fixed flat port, it appears that, of the options discussed above, the most reliable way to produce wide-angle underwater pictures which do not require computer correction is to use a converter designed for 35mm lenses and zoom in to eliminate the vignette. One thing we have not produced however, is any evidence that the adapters designed for 28mm are at fault. It is just, perhaps, that they pursue a goal (no vignetting) which is not necessarily compatible with best optical quality.
The problem of aberration, of course, evaporates if radial correction is carried out as a matter of routine. The use of Panorama Tools in Photoshop is probably a bit too technical for most people, but lens correction with Adobe Lightroom is easy (and the program provides RAW converters for a large range of cameras).
An important consideration which seems to get lost in the general discussion of wide adapters is that a camera which can zoom out to f = 28 or even 24mm can perfectly well be used at f = 35mm. The Canon S100, for example, can even be set up so that specific focal lengths (24, 28, 35, etc.) can be selected using the front control dial. This means that, provided that the camera can reach (or go to less than) the adapter's target focal length, then the adapter can always reach its target angle of coverage. Thus, apart from the convenience of using the extreme wide zoom setting as a point of reference, nothing is lost by having to zoom-in until vignetting disappears.
See also: Computer lens correction.
© Cameras Underwater Ltd. 2012
David Knight asserts the right to be recognised as author of this work.