Ultra Wide Angle Lenses on Digital SLRs
MYTH: Wide angle lenses on digital SLRs lead to weird chromatic problems at the corners of the image since there is an offset between the light-gathering microlenses of the CCD and the CCD itself. Therefore a myth was created (I think by Olympus who made no digital SLRs at the time) to scare people into thinking that using wide angle lenses on these cameras would lead to nasty color fringes at the corners due to the sharp angles of incidence of the light beams. Supposedly since the light was not perpendicular to the CCD the light would pass through the microlenses at an angle and hit the wrong pixels.
Having spread this myth, some impressionable people thought they needed new magic lenses that are supposed to have light coming out of the back to hit CCDs magically at 90 degree angles.
FACT: This is completely irrelevant. This is just FUD created by a camera company that did not make real digital SLRs as well as makers of discount lenses. "FUD" is an acronym used by salesmen and marketers meaning "Fear, Uncertainty and Doubt." FUD campaigns are waged by salesmen and companies that cannot or do not make a certain kind of product in the hope that their potential customers will be discouraged from buying what they really want and instead buy something made by the party dishing out the FUD.
This clever myth plays on most people's lack of familiarity with the specifics of lens design. All wide angle SLR lenses have been designed to eliminate this for the past 50 years. The retrofocus design, invented at the French company Angeneiux in the 1950s allows wide angle lenses to be used on SLRs by moving the rear exit pupil forward to clear the flipping mirror of the SLR. Otherwise the back of the short focal length lens would be so close to the film that the mirror would get caught. If one used a traditional short focal length lens you might have these chromatic issues, however on SLRs the lens designs are always retrofocus and therefore this has never been a problem.
An exit pupil is an imaginary point through which light rays seem to pass. A lens has two of these, front (entrance pupil) and rear. When you look in the front of a lens you are looking through the front entrance pupil, also called the front nodal point. There will be a physical point around which you can rotate the lens to the left and right and the diaphragm of the lens will appear to stay put. Locate this point and you have located the front nodal point of your lens. I other words, to locate this point wiggle your lens around different points until you discover the point from which you can wiggle the lens while the diaphragm seems to stay put. Likewise, looking through the rear of a lens you can locate the rear exit pupil point.
The rear nodal point is the point from which light appears to come. In SLR wide angle lenses, regardless of effective focal length, the rear exit pupil is still about two inches from the film or CCD. Camera makers also do this to optimize the viewfinder system for their lenses, another reason you can avoid no-name discount lenses which are not optimized for your camera the way camera-brand lenses are.
Very simple lenses, like magnifying glasses, have these two nodal points in the same place.
Normal SLR lenses and conventional lenses for view cameras also tend to have the two nodal points close together.
Wide angle SLR lenses and telephoto lenses spread these apart deliberately.
Wide angle lenses put the front nodal point further away from the film than the actual focal length of the lens. These were invented to fix the problem of interfering with the mirror of an SLR and thus this has never been a problem. This retrofocus gyration is why wide angle lenses for view cameras and rangefinder cameras have been better optically than SLR lenses, however today with digital SLRs they have that problem fixed before it even started.
Telephoto lenses move the front nodal point closer to the film than the focal length of the lens, allowing compact telephotos. Get out a ruler and you'll discover your 300mm lens does not poke out 300mm (one foot) from your camera.
Guess what else: due to all this most SLR lenses have their rear nodal points in about the same place regardless of focal length, and SLR viewing and metering systems are designed around all this. This is why one needs to meter with a shift lens before shifting it, and why you can forget the old wives' tale about this being a problem with digital SLRs. The cameras are intended to work with lenses having the light come from the rear nodal point that it does.
In fact, since the rear exit pupils of most SLR lenses of the same brand are at about the same distance from the film plane or CCD one could design one's cameras with the microlenses positioned exactly for this, and for all we know, they may be designed exactly this way. In this case, the stupidest thing you could do is buy a Sigma lens claiming to have magic parallel light beams coming out the back (rear nodal point at infinity) and completely screw up the design of your Nikon camera. Hey, no one pays me anything for anything, it's just sad when people are hoodwinked by salespeople into buying crud for exactly the wrong reason.
So why do some people see color fringing in the corners?
Simple: some lenses have visible lateral chromatic aberration, and some digital cameras have enough resolution to make this clearer than it can be on film if you look for it. This effect has always been with us and I see it when I look for it on film. The Canon 14mm lens has never been very good (see Photodo.com for instance) and therefore putting it on a Canon 1Ds you'll see this. Shoot with the superior Nikon 17-35 as I do and it's not a problem.