Idaho Airships, Inc.

There is an old adage that, “If you haven’t made any enemies, you haven’t done anything.” In this post, I plan to do plenty.

I have encountered many professional shooters, some of whom I would classify as “elite” photographers, who believe that using a camera with a crop factor somehow increases the effective focal length (”reach”) of their lenses. Not true.

There are some related points of contention that we should address before stampeding off to our conclusion:

• Sensor placement is immaterial in this matter because of the relatively limited amount of variance that can be designed into 35mm format DSLR’s.
• “Telephoto” most appropriately refers to lenses that are physically shorter than their focal length. The correct term for “long” lenses (those exceeding the focal distance of the normal-itself an optical science term not widely understood, which in the 35mm format is about 50mm)-is “narrow angle lens.” This is according to Dr. Rudolf Kingslake, former Director of Optical Design for Kodak, Emeritus Professor of the University of Rochester: Optics in Photography, published by the International Society for Optical Engineering, 1992, pp. 49-50.

The size of a sensor has absolutely no effect on the lens. Decreasing sensor size does not increase the magnification of the lens, reduce aberrations, or increase the depth of field (DOF).

All the size of the smaller sensor does is reduce the area of the image captured by the camera. Nothing more.

All of the small-sensor effects being touted, such as somehow transmogrifying a 400mm lens into a 520mm or 640mm lens, cropping out the poorer-performing periphery of a lens, increasing DOF, and relaxing the need to carry very large, heavy lenses…are actually logistic matters introduced by lens selection.

An exhaustive explanation of this paradox far exceeds a blog post, but we could construct 30 representative shoot assignments and configure, for example, a 5D (full frame) and a 20D (1.6x) for each, and it would be easy to show the functional superiority of the full frame, particularly in tight positions-the full frame image can always be electively cropped in post processing.

• Dr. Roger N. Clark explains the DOF myth.
• Paul van Walree corroborates the DOF myth (see #7).
• Sean T. McHugh, Ph. D. explains other issues of sensor size.

We are left then with only the question of capturing higher resolution at the sensor. It is true that the smaller sensors have smaller photosites, and thus smaller pixels following Bayer Array Demosaicing. Our example 20D has a pixel pitch of 6.4u (6.4 microns, or 6.4 millionths of a meter), and the 5D has a pixel pitch of 8.2u. For comparison, a “high resolution” crop factor camera, the Nikon D2x, has a pixel pitch of 5.5u. This can obviously generate questions concerning the ability of smaller pixels being able to capture more detail in a give space. That question is very reasonable, but the answer is convoluted and frustrating.

All other things being equal (sensor quality, system configuration, exposure, etc.), larger photosites produce better images. They capture more photons. Of course, noise is only a fraction of the issue, but it is a significant fraction. Noise most certainly affects image quality: Color and contrast at the pixel/detail level.

The remainder of the reasons that the smaller pixels are a liability involve esoteric optical science. In summary, Diffraction Limiting is the primary culprit, but Seidel Aberrations, Chromatic Aberrations, and Modulation Transfer irregularities make major contributions.

• Paul van Walree encapsulates optical degredations.
• Sean T. McHugh, Ph. D. explains Diffraction Limiting and Airy Disks.
• Norman Koren explains Modulation Transfer Function (the imperfect transmission & collection of visible band light).

The decisive point in this matter is that all of these optical degradations are additive and render pixel pitches of less than about 8.5-9u moot except in extremely rare and controlled circumstances (such as in a studio). All but a handful of Canon’s L Series lenses simply can’t produce a sensor spot size of less than about 7-8u except in atypical configurations, and the matter is only exacerbated by hand holding the camera-a common practice in aerial photography. Thus, the lesser pixel pitch of the smaller sensors simply has little or no opportunity to produce detail beyond the pixel pitch of the full frame 5D…which itself may be superfluous but does produce higher quality pixels.

There are many more questions to answer relative to this matter. Most photographers notice, for example, that when they view their images under high magnification, the various pixels seem fully resolved because they are distinguishable or different than their neighboring pixels. Being so doesn’t indicate bona fide resolution, since each pixel is a calculated entity and profoundly affected by sensors on all sides, meaning that there is no way, short of disassembling the Bayer Demosaic, to tell how much each sensor site contributed to each particular pixel, or how much noise contributed to the final pixel and its difference from its neighbors.

These issues will be further developed in a subsequent post comparing the 5D to the 1DsMII-a comparison that has already generated a bit of heat when it was distributed around the Professional Aerial Photographers Association in early 2007.

L