Why aren't lenses designed with extra large apertures for autofocus beyond what's used for imaging?

Question

Historically, designing large-aperture lenses has been difficult because correcting the optical aberrations that arise from large apertures requires complex designs and very large amounts of glass. Theoretically, if one didn't care about aberrations, it would be easy to design a lens with nearly any arbitrary aperture. In such a case, the only difference between say, an f/1.4 and an f/4 design would be the physical size of the aperture, and therefore of the front element.

Given that this is the case, it should be relatively trivial to design a lens that always stopped down to an acceptable aperture to take images, but remained at a small f-stop when focusing. For example, if we were to take a hypothetical 200mm f/4 lens, the lens would only take images at f/4 or narrower apertures, but it might focus at say, f/2.8 or even f/2.

The additional amount of glass in such a design only needs to be just enough to correct the optical aberrations for sufficiently precise focus to be achieved and to let enough light in, allowing one to save weight and material and therefore cut down on price. The design should, in the best case, be essentially identical to a lens with the same "true" aperture as the "photo-taking" aperture of the hypothetical lens, apart from the larger front element.

I'm referring specifically to lens designs that don't exhibit significant focus shift when stopping down. Obviously, lens designs that have sufficient spherical abberations for this to become a problem could not be designed in this way.

Such a design would benefit lenses at all price ranges, since it would make it possible to make faster-focusing consumer zooms at only slightly higher prices but also to make fast-focusing premium extreme telephotos (such as an 800mm lens that focuses at f/4 but shoots at f/5.6).

Is there a technical or commercial reason why such designs aren't in widespread use?

Answer 1

The fundamental driver of cost in a lens is not the correction of aberrations, although the correction of aberrations does add to the cost of a lens, and may be a more significant factor in wider angle lenses. Generally speaking, the primary cost of a lens is the "glass". I put glass in quotes, because sometimes it is other materials, such as Fluorite or a diffraction grating or diffractive particle dispersion, however advanced lens elements usually cost MORE.

You cannot achieve a specified aperture without having the appropriate magnification from both ends of the lens for that aperture to appear to be the correct size. The notion of a "physical aperture" is generally a misnomer. What we call the aperture of a lens, what is frequently referred to as the physical aperture, is correctly termed the entrance pupil. The entrance pupil is the aperture as observed through the front of the lens at a distance of "infinity" (or, in other words, a sufficiently great distance that the observation is of collimated light.) The entrance pupil of a 600mm lens with an f/4 relative aperture must be 150mm as observed through the front of the lens. To achieve that magnification, two things must be:

1. The right lens elements each with the right magnification must be used to achieve that magnification.
2. The front element must be at least 150mm in diameter.

Think about this for a moment...a 150mm diameter front lens element. That is 6" in diameter, about a hands breadth. That is HUGE. On top of that, the front half of the barrel up to the diaphragm is only slightly tapered, and there are a number of additional lens elements that must be used in addition to the front element to achieve the point #1 as well as correct for aberrations. So you have a number of 4" to 6" lens element in the front half of the barrel, on top of all the lens elements behind the diaphragm to properly project a rectilinear image onto the sensor, each of which are still an inch to several inches in diameter. All of that glass COSTS!

Replacing an aspherical element that corrects for aberrations with a spherical element that does not will probably reduce cost, but not by a particularly significant amount unless we are talking about very short focal lengths are wide angle zooms (where the cost of aberration corrections tends to be a more significant cost factor, as overall glass quantity is much lower than in longer focal lengths.) Regardless of the type of lens, however, usually the larger portion of cost is the front element, maybe the front couple of elements.

Even in a wide angle lens, the front element will usually be many times larger than necessary to achieve the correct entrance pupil, only in this case it is necessary to bend light from a sufficiently wide angle, rather than to gather the required amount of light. In a wide angle lens, the front element can be many times the total volume of any other single lens element. Glass costs.

As you state, the amount of additional glass need only be enough to allow for the larger aperture. Remember that every stop in aperture is a FACTOR OF TWO change in aperture AREA. If you have a 600mm f/5.6 lens, the aperture is 107mm, or ~11,500mm^2. If you want to make a 600mm f/4 lens, even if you don't care about correcting aberrations, that is a 150mm aperture...or 22,500mm^2!! You've DOUBLED the minimum area required to support the aperture, and likely more than doubled the volume (larger elements are usually thicker as well, so the total increase in volume can more than double). And that is just for the front element...there are still around 12-18 more! The amount of glass, in terms of volume, required to just barely be enough to support the larger aperture is more than double what is necessary for the next stop down. Don't underestimate that cost.

As others have mentioned, a lot of lenses already do exactly what you have theorized: Allow IQ to suffer at maximum aperture, requiring the lens to be stopped down by as much as a stop or so to fully realize maximum sharpness potential. Generally speaking, cheaper consumer grade lenses do such, usually as a result of a multitude of factors (cheaper glass, simpler and fully automated manufacturing process, automated assembly, etc.)

The single primary reason why cheaper lenses are cheaper, however...is smaller maximum apertures. Most consumer-grade lenses, as well as most third-party lenses, use smaller maximum apertures. Most consumer-grade wide angle and telephoto zooms max out at f/3.5, usually non-constant so f/3.5-5.6. Many third-party telephoto zooms start at f/5.6 while brand names start at f/4, and third-party telephoto zooms will often use f/5.6-6.3 while brand-name telephoto zooms often offer f/4 or even f/2.8 constant aperture. Maximum aperture is the primary driver of cost, as it explicitly drives the total volume of glass required.

Corrective lens elements designed to reduce aberrations, such as aspheric elements, Fluorite elements, diffractive elements, ultra-low dispersion elements, etc. all add to cost, however again making a large 5" fluorite element for an f/4 telephoto lens is considerably more costly than making a 3" fluorite element for an f/5.6 telephoto lens. Again...every stop change in aperture is a factor of two change in area, and an even greater change in the total volume of glass.

Answer 2

There's one teensy-weensy little flaw in the plan: the aberrations (particularly, but not solely, spherical aberration) are the main culprit for the phenomenon of focus shift. Essentially, that would mean that your "focusing lens" won't have exactly the same focal length as your "taking lens", so images focused perfecctly at the larger aperture will be out of focus at smaller apertures (at least until depth of field grows large enough to cover/mask the difference).

There are ways around the problem, of course. Hasselblad maps the known focus shifts of its lenses at various apertures as part of its True Focus system. But to get there, and do it reliably, required Hasselblad to close the system. (As well as to marry bodies and backs. Don't worry too much about that; the system allows for a complex polygamous/mate-swapping society, provided that all of the hookups are properly blessed at the factory.) In the free-for-all world of APS-C/135-format DSLRs and EVIL/MILC cameras, that would mean that the new lens type would only work on new bodies, and that the bodies would have to "know" the characteristics of the lenses using that strategy. And it would mean even tighter tolerances than are currently allowed (and we need microfocus adjstments to make them work at a high level) so that the predicted focus shift and the actual focus shift correspond well.

So you save a little money on one part of the lens (the glass) by making both the rest of the lens and the camera body more expensive. And you have to explain to people why their obviously f/1.2 lenses are really only f/2.8s, which would take a bit of marketing. And you'd likely be cutting off the third-party blood supply. That may be fine if you're Hasselblad, simplifying what had been a Frankenstein experience (it's really the only interchangeable-back MF digital that doesn't feel like random parts stuck together at the moment), are in a limited and rarified market, and only have one competitor. It's a huge risk with a lot of potential down-sides in the "commodity" market.

Answer 3

You've probably heard people describe some lenses as "unusably soft wide open, but passable at f/2.8 and excellent from f/4", or similar. That's because, basically, these lenses are already designed in the way you suggest, although additionally constrained by size, weight, complexity, cost, and other design factors. And they also let you use the lens at wider apertures if you reall want to take a picture that way, as a sort of bonus.

In addition to the basic problem of focus shift, more lenses don't do this in an even more extreme way because in the real world those other factors are very significant — and, generally, more important than this benefit. Increasing the maximum aperture, even without regard to image quality requires a lens to be bigger, heavier, and more expensive, just because glass is all of those things. Few people would want to make that those tradeoffs in return for slightly better autofocus.

Answer 4

The extra glass in a fast lens is not just there to correct aberrations. The full aperture must be visible across the whole field of view meaning for moderate or wide lenses, you can't just make the aperture larger you'd have to make all elements in front of the aperture much larger as well.

But your idea is sort of in effect with large format lenses. Many of these lenses are f/5.6, but that's just to prove a bright enough image to focus using the ground glass, you would shoot at this aperture due to depth of field, corner and sharpness issues.