Benjamin Marks
Mentor
Folks: Every lens review I have ever read has noted contrast improving and aberrations decreasing as a lens is stopped down. I have experienced this myself, and my general practice is to ignore it when I shoot. If I need f:1.4 to make a shot work, I need it. And if I have chosen f:1.4, I have chosen it because either I need it to maintain a minimum shutter speed, or I need it to isolate a subject.
But this set of stop-down effects exists whether the lens in question is an f:1.4 lens or an f:2.8 lens, and that has always confused me. Formal performance is always worst when a lens is wide open, not that it may matter (as I note above) in practice. But why not over design your lens so that, for instance, at f:2 your lens is already "stopped down" and exhibiting superior performance.
I am sure that there is something simple and fundamental that I am missing here. Perhaps the nature of optics is such that an f:2 lens and an f:0.95 lens both exhibit the same amount of coma, say, at f:2, and that f:2 in the faster lens simply gives better performance in a comparative sense than the the same lens does wide open.
But if this were the case, why do slow f:4 or f:5.6 zooms not perform optimally wide open?
But this set of stop-down effects exists whether the lens in question is an f:1.4 lens or an f:2.8 lens, and that has always confused me. Formal performance is always worst when a lens is wide open, not that it may matter (as I note above) in practice. But why not over design your lens so that, for instance, at f:2 your lens is already "stopped down" and exhibiting superior performance.
I am sure that there is something simple and fundamental that I am missing here. Perhaps the nature of optics is such that an f:2 lens and an f:0.95 lens both exhibit the same amount of coma, say, at f:2, and that f:2 in the faster lens simply gives better performance in a comparative sense than the the same lens does wide open.
But if this were the case, why do slow f:4 or f:5.6 zooms not perform optimally wide open?
zuiko85
Mentor
I believe that if a lens was designed to operate at only one aperture, such as the 15mm f3.5 on Minox 8X11 cameras, then it can be optimized for wide open.
However if it is to operate over a range of apertures then there are compromises that are inherently unavoidable.
Kind of like standard angle lenses that are corrected for regular focusing range, say 3ft. to infinity, may not do as well in the close focusing range of 1:1 to 1:10 reproduction ranges.
A swiss army knife may have a screwdriver blade, but a properly sized screwdriver will still be a better tool.......if it is available.
However if it is to operate over a range of apertures then there are compromises that are inherently unavoidable.
Kind of like standard angle lenses that are corrected for regular focusing range, say 3ft. to infinity, may not do as well in the close focusing range of 1:1 to 1:10 reproduction ranges.
A swiss army knife may have a screwdriver blade, but a properly sized screwdriver will still be a better tool.......if it is available.
astrosecret
Recovering rollei snob
i guess i don't really understand the question. a faster lens needs more glass to collect more light. depending on maximum aperture the glass is formed shaped and designed a specific way to project a path of light properly. as such it will perform a specific way. When a lens closes down, you eventually run into diffraction.
look at a lens. 1.4 lens has more glass than f2. the glass inside is differently proportioned physically and will perform differently, and people can manipulate it a little bit and some designs are improvements upon others. that is why many lenses that have an f2 max perform a lot better at f2 than some lenses with a max 1.4 or 1--it's just better designed. but nothing defies the laws of physics. every single lens with multiple apertures no matter how well designed will perform differently at different apertures... whether one aperture or another performs 'badly' is a matter of taste i guess.
look at a lens. 1.4 lens has more glass than f2. the glass inside is differently proportioned physically and will perform differently, and people can manipulate it a little bit and some designs are improvements upon others. that is why many lenses that have an f2 max perform a lot better at f2 than some lenses with a max 1.4 or 1--it's just better designed. but nothing defies the laws of physics. every single lens with multiple apertures no matter how well designed will perform differently at different apertures... whether one aperture or another performs 'badly' is a matter of taste i guess.
Rayt
Nonplayer Character
Expensive lenses tend to perform better at max aperture. Is that what you are referring to?
over-design greatly increases the cost.
sebastel
coarse art umbrascriptor
and the weight and the size.
yes, such a lens would need a considerably larger image circle
astrosecret
Recovering rollei snob
at a certain point wont you just be manipulating the projection/capture surface to capture the ‘best’ part if the lens. and best is subjective... many people want an image to have a certain character that are properties of lenses at different apertures
KoNickon
Nick Merritt
What I have gleaned (which may well be incorrect):
The wider the aperture, the more aberrations exist. Wider-aperture lenses require more elements to counteract those aberrations. And as I understand it, the smaller the maximum aperture, the less difference will be between maximum and smaller apertures. So a Tessar-type design works quite well at maximum aperture, if the maximum aperture is f3.5 or smaller, as I understand it. And the difference between maximum aperture and two stops down is less likely to exhibit differences in quality as compared with a faster (f2.0, say) lens at full aperture vs. two stops down.
The OP's question, in part, was as I understand it asking, "So why not just design a Sonnar or Summicron (for instance) with a smaller maximum aperture?" Probably that will give you a lens with very even performance from maximum aperture down to a much smaller aperture, but as others have pointed out, that's going to make for a lens that's much more complicated (and more expensive) than it needs to be. The benefits of the more complex design -- which is intended to address the wider-aperture aberrations -- are going to be lost.
The wider the aperture, the more aberrations exist. Wider-aperture lenses require more elements to counteract those aberrations. And as I understand it, the smaller the maximum aperture, the less difference will be between maximum and smaller apertures. So a Tessar-type design works quite well at maximum aperture, if the maximum aperture is f3.5 or smaller, as I understand it. And the difference between maximum aperture and two stops down is less likely to exhibit differences in quality as compared with a faster (f2.0, say) lens at full aperture vs. two stops down.
The OP's question, in part, was as I understand it asking, "So why not just design a Sonnar or Summicron (for instance) with a smaller maximum aperture?" Probably that will give you a lens with very even performance from maximum aperture down to a much smaller aperture, but as others have pointed out, that's going to make for a lens that's much more complicated (and more expensive) than it needs to be. The benefits of the more complex design -- which is intended to address the wider-aperture aberrations -- are going to be lost.
Godfrey
somewhat colored
If you design a lens to operate at ONLY one lens opening, you can optimize it fully (to within the limits of the technology and materials available) to give its best performance at that opening over some given range of focus distances. If you design a lens to provide a choice of lens openings, you have to make some compromises about when it achieves its optimum performance and over what range of focus distances.
Why?
Because a lens is made of various kinds of glass held at specific distance relationships to direct the light moving through it. The different corrections needed for different lens openings and different focus distances imply changes to the lens design required for optimization: glass cannot change shape as the aperture and focus distance settings are varied to compensate. "Floating element" lenses are computed to try to make the compromises required smaller compromises, but the glass itself cannot change shape and there is a limit—with the helical threaded lens mountings and focus distance adjustments and the various kinds of glasses and cements and edge blacking able to be used—to how much accommodation can be managed given the nature of the materials and the cost of manufacture.
What you see when you look at competing max-aperture lenses (for instance, a 75mm f/4 vs a 75mm f/1.4 both designed for full-frame 35mm format), is that the typical change in lens optimization from wide open to two-three stops down from wide open is smaller for the 'slow' lens than it is for the 'fast' lens. This is because the slow lens' smaller glass surfaces can be computed with fewer compromises and manufactured to slightly smaller variances than the fast lens, if manufacturing technology and price are held constant, and also because the relative sizes of the lens openings and glass surfaces involved are smaller as well.
G
"There are no dumb questions. In the absence of understanding, not asking questions is far dumber than any question at all."
benlees
Well-known
Consumer grade lenses are compromises based on price points and function, determined by the physical properties of light hitting a focal plane. On the other end of your example is diffraction. If you use f22 on your small sensor camera you are sacrificing image quality, but if you want the depth of field then maybe that's a trade off that works for you. At f2 the diameter of lens opening is so wide that an average lens can't direct the light with the accuracy of a more stopped down position. Most consumers (I'm guessing) prefer some depth of field so the design of lens optimizes the low mid apertures, which also avoids competing with diffraction at the small sized apertures. A $100 kit lens is no match for a $4000 Zeiss. Whether or not someone can tell by looking at actual prints is another matter!
retinax
Well-known
If I understand your question right, it could be re-stated as follows: Why aren't slower lenses made in such a way that they perform as good wide open as a faster lens would, stopped down to the same aperture?
First, that is the case very often, but there are exceptions. Especially in the past. E.g. the early very fast lenses for Leica have a reputation of being much weaker than the slower ones.
Second, I think the answer is mostly one of economics. It's perfectly possible, but would mean the slower lens would have to be optically rather complex. So it would cost nearly as much as the faster lens. How many people would buy the slower lens then? So manufacturers chose simpler formulas for slower lenses, so that they can be offered significantly cheaper. The slow zoom you mentioned wouldn't be much cheaper and smaller than a very fast zoom if it was made to be diffraction limited wide open.
Another consideration is that with SLRs, the wide apertures are helpful for viewing and focusing, so there's a justification for them being available, even if they're not good enough for picture taking except special occasions.
First, that is the case very often, but there are exceptions. Especially in the past. E.g. the early very fast lenses for Leica have a reputation of being much weaker than the slower ones.
Second, I think the answer is mostly one of economics. It's perfectly possible, but would mean the slower lens would have to be optically rather complex. So it would cost nearly as much as the faster lens. How many people would buy the slower lens then? So manufacturers chose simpler formulas for slower lenses, so that they can be offered significantly cheaper. The slow zoom you mentioned wouldn't be much cheaper and smaller than a very fast zoom if it was made to be diffraction limited wide open.
Another consideration is that with SLRs, the wide apertures are helpful for viewing and focusing, so there's a justification for them being available, even if they're not good enough for picture taking except special occasions.
Both the Leica Summar and Summitar are F2 lenses, the Summitar is the better performer. The Summar was known for severe vignetting when used wide-open. With color film becoming more popular, the deficiency was more noticeable. In the "Leica Manual" by Morgan and Lester, attention is brought to the front element of the Summitar, described as appropriate for a lens of an F1.6 aperture. In effect, the designers correct vignetting of the Summar by designing a faster lens that was used at F2.
In the Kodak Retina series, you have the IIIC with a 50mm F2 Xenon, the IIC with a 50mm F2.8 Xenon, and the IB with the 50mm F2.8 Xenar, a Tessar type lens. The rear group of the IIIC and IIC are identical. The front group is essentially the F2 lens in a smaller housing with the edges clipped. Replace the front element- the IIB works at F2. The F2.8 Xenon is a much better performer than the Xenar.
Some of the better performing lenses are faster designs that have been "stopped down" as built.
In the Kodak Retina series, you have the IIIC with a 50mm F2 Xenon, the IIC with a 50mm F2.8 Xenon, and the IB with the 50mm F2.8 Xenar, a Tessar type lens. The rear group of the IIIC and IIC are identical. The front group is essentially the F2 lens in a smaller housing with the edges clipped. Replace the front element- the IIB works at F2. The F2.8 Xenon is a much better performer than the Xenar.
Some of the better performing lenses are faster designs that have been "stopped down" as built.
aizan
Mentor
Lenses can be designed for peak performance at open aperture. For example, smartphone cameras and microscopes. But that's only practical because the lenses are small. If you wanted to do that on a regular-sized camera, the lens would be massive.
zuiko85
Mentor
I have a Olympus 38mm 6 element f2.8 enlarging lens and there is a permanent stop built into the lens. I’ve often wondered if it is actually a tweaked 38mm f1.8 lens mounted in a non focusing barrel, but have no way to confirm this suspicion.
peterm1
Mentor
"Formal performance is always worst when a lens is wide open, not that it may matter (as I note above) in practice. But why not over design your lens so that, for instance, at f:2 your lens is already "stopped down" and exhibiting superior performance."
Not a dumb question at all. Well, you could perhaps. For example by having lens elements of sufficiently large size you could in principle relatively easily design a lens such that it would even at the largest of open apertures be using only the centre portion of the lens - it effectively would be shooting stopped down all the time.
Some (perhaps many) lenses already seem to be working on the principle you suggest. If one looks for example at some of the highly corrected lenses coming out of Sigma's design shop they are huge. Compare say, a Nikkor 50mm f1.4 AF D (or the later equivalents) with a Sigma Art 50mm f1.4. The Nikkor is about 230 grams and the Sigma about 820 grams give or take. The Sigma performs brilliantly but this is the cost you pay for it (as is too, the actual dollar cost - it is far from being a cheap "normal" lens). Lens design is always a compromise.
Oh by the way another cost comes to mind too. For people like me who value some degree of "character" in a lens for artistic purposes I suspect the above approach reduces that opportunity. At least some of that "character" comes when shooting at wider apertures from residual aberrations inherent in the lens design. And that is usually more evident at those wider apertures from the fact that those aberrations are less well able to be corrected towards the outer part of the image unless you have exotic lens designs. This is one reason aspherical lens elements were implemented - to overcome, for example, spherical aberration due to the fact that spherical elements can less well correct for it especially in the outer parts of those elements. Another way of correcting it is to have honking great big elements (as you suggest) and only using the centre portions of them for all your shooting. Or of course simply by stopping down.
Not a dumb question at all. Well, you could perhaps. For example by having lens elements of sufficiently large size you could in principle relatively easily design a lens such that it would even at the largest of open apertures be using only the centre portion of the lens - it effectively would be shooting stopped down all the time.
Some (perhaps many) lenses already seem to be working on the principle you suggest. If one looks for example at some of the highly corrected lenses coming out of Sigma's design shop they are huge. Compare say, a Nikkor 50mm f1.4 AF D (or the later equivalents) with a Sigma Art 50mm f1.4. The Nikkor is about 230 grams and the Sigma about 820 grams give or take. The Sigma performs brilliantly but this is the cost you pay for it (as is too, the actual dollar cost - it is far from being a cheap "normal" lens). Lens design is always a compromise.
Oh by the way another cost comes to mind too. For people like me who value some degree of "character" in a lens for artistic purposes I suspect the above approach reduces that opportunity. At least some of that "character" comes when shooting at wider apertures from residual aberrations inherent in the lens design. And that is usually more evident at those wider apertures from the fact that those aberrations are less well able to be corrected towards the outer part of the image unless you have exotic lens designs. This is one reason aspherical lens elements were implemented - to overcome, for example, spherical aberration due to the fact that spherical elements can less well correct for it especially in the outer parts of those elements. Another way of correcting it is to have honking great big elements (as you suggest) and only using the centre portions of them for all your shooting. Or of course simply by stopping down.
Benjamin Marks
Mentor
Thank you all for your thoughtful responses. They have taught me much. Funny, I was organizing some lenses this weekend and handled a 35mm Distagon for a Nikon. That thing is huge compared to the Summicron-M. Part of that is RF vs. SLR. But the Distagon is also perhaps four times larger than the 35/2 Nikkor, as well. Many thanks to all who answered.
GLL
Member
I am definitely no optics expert but to add to the discussion - like others in the thread it seems to be that optical design is effectively a trade off between a number of different factors. Putting aside more subjective elements like 'character', it seems like size of the lens, speed, flatness of field, correction of aberrations and image circle are all factors that lens designers have to trade off against each other.
Others have already given examples of lens designs which are effectively 'stopped down' versions of other lenses. I would also point to lenses which appear to have traded speed directly for size (like the voigtlander color skopar lenses), or vice-versa, have traded size for speed. The 7Artisans 50 1.1 is an interesting example - it's a large, beer can of a lens that apparently can cover the full image circle of the Fuji GFX (with some vignetting in the corners). This suggests to me that increasing the image circle can be a bit of an optical design hack to increase the performance of a lens in the corners in exchange for making it large and chunky.
Another interesting lens in this space is the VC 50 3.5 Heliar, which is both very compact and allegedly very sharp (though i've never seen it tested head to head with a 'cron on an optical bench). I've heard people mention that Heliars are very difficult to design at faster speeds - I wonder if they effectively 'designed it stopped down'.
Others have already given examples of lens designs which are effectively 'stopped down' versions of other lenses. I would also point to lenses which appear to have traded speed directly for size (like the voigtlander color skopar lenses), or vice-versa, have traded size for speed. The 7Artisans 50 1.1 is an interesting example - it's a large, beer can of a lens that apparently can cover the full image circle of the Fuji GFX (with some vignetting in the corners). This suggests to me that increasing the image circle can be a bit of an optical design hack to increase the performance of a lens in the corners in exchange for making it large and chunky.
Another interesting lens in this space is the VC 50 3.5 Heliar, which is both very compact and allegedly very sharp (though i've never seen it tested head to head with a 'cron on an optical bench). I've heard people mention that Heliars are very difficult to design at faster speeds - I wonder if they effectively 'designed it stopped down'.
markjwyatt
Well-known
f-stop = focal length/physical-aperture
If you measure the physical aperture (sight through the lens with caliper jaws between your eye and the outer lens- back or front, and close the caliper until you are tangent on two sides of the aperture- you may have to fudge it depending on shape), you can confirm approx. f-stop is 38mm/(measurement of physical aperture in mm). Try it and see if your get something closer to 2 than 3, and you should be able to figure it out. Do this in a way that you avoid scratching your lens! f2.8 should have a 13.57mm dia aperture (close to 17/32", roughly a bit over 1/2"); f1.8=21.1mm (between 3/4 and 7/8", or about 53/64").
I never tried it, but it should work.
pvdhaar
Peter
It's a consequence of physics, for which there is no perfect technological solution.
The breaking index of glass depends on the wavelength of the light (think prisma), and the amount that the path of the light needs to be broken increases with the distance from the center of the lens. The result is that the wider the lens is, the bigger the difference is where light of different wavelengths hits the focal plane.
This effect is typically mitigated by optimizing lens design and glass choice for 3 wavelengths (e.g. red/green/blue) for a specific focus distance (e.g. 10ft/3m), but even though our eyes (and digital sensors and flim for that matter) capture only a limited number of discrete wavelengths, the real world exhibits light with a continuum of wavelengths. You never get perfect focus over the entire spectrum for light hitting the lens over its entire surface.
When you stop a lens down, light only hits the lens closer to the center, and needs to be broken less, with the result that different wavelengths coincide more on the focal plane.
The breaking index of glass depends on the wavelength of the light (think prisma), and the amount that the path of the light needs to be broken increases with the distance from the center of the lens. The result is that the wider the lens is, the bigger the difference is where light of different wavelengths hits the focal plane.
This effect is typically mitigated by optimizing lens design and glass choice for 3 wavelengths (e.g. red/green/blue) for a specific focus distance (e.g. 10ft/3m), but even though our eyes (and digital sensors and flim for that matter) capture only a limited number of discrete wavelengths, the real world exhibits light with a continuum of wavelengths. You never get perfect focus over the entire spectrum for light hitting the lens over its entire surface.
When you stop a lens down, light only hits the lens closer to the center, and needs to be broken less, with the result that different wavelengths coincide more on the focal plane.