Does the imaging circle size change when a lens zooms?

Question

When a zoom lens is designed for a cropped (APS-C) imaging circle, does the imaging circle necessarily stay that small throughout the zoom range?

It is easily understandable that a prime lens designed for an APS-C camera, like the Nikkor DX 35mm F/1.8G, would have a smaller imaging circle than a full-frame lens such as the Nikkor 35mm F/1.4G.

However, how does this work for zoom lenses? Take the Nikkor DX 16-85mm for example. It is a DX lens designed for APS-C camera but when you zoom past 24mm (16 times the 1.5X crop), wouldn't the imaging circle get big enough for a full-frame sensor? Or not? Why?

Finally, what would be the angle-of-view of the DX zoom at 24mm on a full-frame camera, if the imaging circle was sufficient to cover the whole sensor area. Currently, many Nikon FX crop the sensor area when a DX lens is mounted, but I am beginning to wonder if they need to do this unconditionally. Maybe they could stop cropping when the lens is zoomed past a certain point?

Answer 1

On some lenses the image circle does indeed get bigger as you zoom.

I know from experience the Canon EF-S 10-22 f/3.5-4.5 does this. At 12mm the image circle is big enough for a 1.3x crop APS-H sized sensor and by 15mm it is big enough for a full frame 35mm sensor.

However lenses such as the EF-S 17-55 f/2.8 vignette at all focal lengths on full frame. I believe the constant max aperture has something to do with it, as otherwise the total amount of light coming out of the lens would somehow change despite the iris and entrance pupil remaining the same size.

Note that the reverse is not necessarily true, the image circle doesn't necessarily get larger with a variable max aperture lens. Some Nikon lenses contain a rear baffle to reduce flare, and this causes the same vignetting regardless of focal length. Integral petal hoods on ultra-wides can have the same effect.

Finally, assuming the DX 16-85mm does not vignette on full frame at 24mm, it will have the same angle of view as any other 24mm lens - 73.7 degrees horizontally and 53.1 degree vertically.

Conditional cropping based on focal length would produce a very strange relationship between focal length and angle of view with the lens getting wider then narrower as you zoom in!

Answer 2

It depends on the individual lens, but it can actually work opposite to the way you're imagining it.

Remember that if the focal length of the lens is less than the sensor-to-back-of-lens-mount distance, then the lens must be of a retrofocus design. (And that with single-lens reflex designs, that distance is going to be something in the neighborhood of 40mm.) That is, the rear portion of the lens is of a much longer focal length than the focal length of the lens system taken as a whole, but it's looking through a concave window on the world. It's sort of like looking through the security peephole in an apartment door through a telescope. (That's why lenses with teeny, tiny focal lengths have such huge front elements—in order for the entrance pupil of the lens to be visible to the entire field of view, the front element needs to be a whole lot bigger than the calculated aperture would suggest.) So at the shortest focal length of the entire lens, the rear segment may actually be configured to be at its longest focal length, and thus have the narrowest light cone behind the lens and the smallest image circle the lens can produce.

There can be any number of ways to achieve the same focal length range, so the pattern of variation in image circle size changes may be different for different zoom lenses of the same range. Browsing through the lens reviews at photozone.de, for instance, and keeping an eye only on the vignetting charts, you'll see that vignetting problems can occur at both ends of the zoom range, or even be worst in the middle of the zoom range. Almost every APS-C/DX zoom, though, displays significant vignetting wide open on the camera it was designed for over most of the range, and that doesn't take into account hard mechanical vignetting outside of the designed image circle, just the slow optical fade. You can always test the image circle off-camera by zooming while projecting the image onto a piece of paper. You're likely to see size changes, but not necessarily where and how you'd predict.

I don't think it would be worth anyone's while, really, to unlock the next crop size up between, say, 42 and 44mm and then again between 58 and 63mm on a given lens because the image circle was slightly bigger in just those ranges—it would just lead to user confusion and consumer complaints.

Answer 3

Simple answer: No, it does not necessarily do so. I used a Sigma 10-20mm f/4-5.6 lens, designed for 1.5x crop cameras, on a Canon 1D (1.3x crop). It was useable - for generous values of useable; the corner sharpness was nothing to write home about and corner vignetting was quite severe - from 12mm up to 20. Below 12mm, the image circle did not cover the sensor.