Optimal f-stop for specific lenses

[dchew]

I'm a bit confused on this topic.  Let's say there are two lenses:
Lens 1.  f/4 lens with a "optimal" f-stop of f/5.6
Lens 2.  f/5.6 lens with an "optimal" f-stop of f/11

I understand a lens wide open has aberrations and other issues that stopping-down improves.  That improvement continues until diffraction issues take over.  I get all that.  But I'll read some posts where people seem to claim lens 2 would be the better lens for those who primarily shoot at f/11.  Here is a recent example:

"One thing I noticed about the Rodenstock HR lenses, especially the wides is they say the ideal aperture is f5.6. Smaller apertures than that can cause diffusion issues. The Apo Sironar digital lenses have from f8-11 as ideal apertures. Even a 35mm lens at f5.6 doesn't have really great DOF, so it would seem that the 35mm Apo Sironar digital 35mm lens might be a more useful lens."
(I kept the poster incognito because I don't want to isolate this poster; I see this kind of statement all over the place - this is just one example)

This is the part I don't understand.  It would seem to me diffraction is diffraction is diffraction, and it always gets worse as the diaphragm opening get's smaller.  Lens 1's sharpest f-stop is f/5.6, but it could still be sharper than lens 2 when both are at f/11.  In other words, just because a lens has an optimal f-stop of f/11, that doesn't tell me anything about which lens is actually "sharper" or "better" at f/11.  The MTF's tell us that.  In fact, if all I knew about these two lenses was the information provided above, I would suspect lens 1 would be sharper at most f-stops including f/11, since for lens 2 it takes two stops for diffraction issues to overcome the lens aberration improvements associated with stopping down.  

Am I missing something?  I always use stupid analogies:  My Jeep gets it's best gas mileage at 50mph.  A Prius get's it's best gas mileage at 40mph.  If I want a car with the best gas mileage at 50mph, I bet it is not my Jeep!

Dave

[henrikfoto]

This is very intersting thoughts. I have been thinking the same.
I would think the sharpest lenses at f4 still in most cases are the sharpest at 8.
F4 with a new 60 or 80 mp back gives very little dof. In many cases you have to close the apperture anyway.

Don't forget the old zeiss optics like the 135mm planar for 4x5 or the Biogon 75mm. These lenses are best at f11 or so and are just as sharp or sharper than the new Rodenstock hr even with digital backs.

[gazwas]

Discounting all other optical issues I would think the difference between the two lenses would be by how much the light must bend in the optical path/design of the lens. The f4 having a larger optical design would need to bend and thus diffract the light more to achieve an aperture of f11 than a lens with an optical design with F5.6 as is maximum aperture.

Both will have to bend and deffract the light to focus the image on the sensor as light travels in straight lines. However when wide open, it will be at a level thats not as noticable as when you stop the lens down further and the light must diffract more for the smaller aperture opening.

Hope i'm not just jibbering.

[Dick Roadnight]

Wide open lenses have more aberrations, and closed down they have more diffraction.

Retro-focus and tele lenses tend to suck, but with good design they can work.

Many lenses are not really designed for taking photos wide open - the max aperture is for viewing and assessing the focus... some lenses, like my WW2 aero-reconnaissance lenses, were designed to be used wide open.

My 900mm is tele-photo, as it had to be to fit in the fuselage of a spitfire!

[dchew]

The f4 having a larger optical design would need to bend and thus diffract the light more to achieve an aperture of f11 than a lens with an optical design with F5.6 as is maximum aperture.

I thought diffraction was associated with the usual photon wave dispersion associated with photons going through a small opening or slit, which is proportional to the size of the opening.  But in my head that is a point source like a laser.  You make an interesting point about the wider angles associated with the larger front element.  My physics is really, really bad...

Dave

[ChristopherBarrett]

I'd think of it this way... take two lenses you're trying to decide between, stop them both down to f/11.  The one with the physically smaller aperture is going to have greater diffraction.  Of course that's only one small element among many factors that make one lens sharper than another.

I've had to take a really different approach to thinking about lenses since I started working with the Red.  I just bought a set of Cooke Panchros for it which have a maximum aperture of 2.9.  They are designed to be shot there and exhibit exceptional sharpness wide open, something I'd never do with Rodenstocks or Schneiders.

In the end I like to just try stuff out and look at the results rather than trying to think about it all and muddle through myriad hypotheses.  Make brain hurt no good.

CB

[dchew]

Thanks Chris,

Actually, I already made my lens decision for now (I went with the Schneider 43xl).  I'm just questioning the statement I see frequently on forum's regarding preferences associated with optimal f-stop for specific lenses, which in general state: If you regularly shoot at f/8 or f/11, get a lens which has it's optimal f-stop in the same range.  I just don't see why...  That "optimum" for a lens is relative only to itself, so really has little to do with lens choice.

BTW, that Chicago PODAS looks enticing. 

Dave

[Bart_van_der_Wolf]

I'd think of it this way... take two lenses you're trying to decide between, stop them both down to f/11.  The one with the physically smaller aperture is going to have greater diffraction.

Hi Chris,

That's not really the relevant issue. F-stops are already relative to focal lengths (magnification factor, optical path length). Therefore diffraction is constant at a given f-stop number, irrespective of the physical aperture size, due to the angular constant of diameter and focal length.

In the end I like to just try stuff out and look at the results rather than trying to think about it all and muddle through myriad hypotheses.  Make brain hurt no good.

Indeed, that's the sensible approach. If needed one can verify/quantify the observations with MTF measurements. Only the relative quality between lenses under comparable conditions matters. Whether one lens is optimal wide open or stopped down, is only important when using that particular lens, and has little bearing on other lenses.

Cheers,
Bart

[teddillard]

The rule of thumb is that a lens performs best at about two stops down from wide-open.  That's been my experience testing lenses, too, what changes is the falloff from there to wide-open, and from there to full closed.  Some lenses are pretty flat- my word- that is the falloff curve doesn't fall off much on either end.  A lens that falls off steeply is going to be less usable.  But the peak always seems to fall around 2 stops down.  

That said, trying to guess at how two lenses compare at various apertures is kind of impossible, you really have to do the tests and run the numbers.  The theory that a faster lens, for example, is going to be sharper at f4 than an f4 lens may, or may not be true...  

(edit: reading through again, I'm not at all sure this addresses your question.  Suffice to say, I've never tested a lens that peaks at the smaller end of the aperture range.)

[leuallen]

Ted

Those are general rules but they don't always hold. Though not MF, I did some lens tests for some of my m4/3 lenses recently and I was surprised. A 100-300 f4.0 zoom was sharper wide open at 135mm than three of my 135mm lenses stopped down from (2.8-3.5) to f4.0. Admittedly, the 135mm lenses were not the best in the world and the zoom is supposedly good.

Also, my Panasonic 100-300 zoom at 200mm is sharpest wide open (5.0) rather than stopped down.

General wisdom with general rules often misleads. I would have thought of my three lenses: Pana 100-300 5.6, Tokina ATX 100-300 4.0, and Nikor 300 ED IF 4.5, the Nikor would win. At least that is what I reached for but not any more. The Panasonic is the winner (at 5.6+), followed by the Tokina, then Nikor.

The only way to know is to test.

Larry

[teddillard]

The only way to know is to test.

Absolutely in agreement.  Thus my use of "rule of thumb" and "my experience".  But that's the point, you just can't come to a conclusion without actual tests...  of the actual lens, for that matter, not a version of the lens.  We found some significant variations from lens to lens within a model, especially with the zooms.

[ondebanks]

Discounting all other optical issues I would think the difference between the two lenses would be by how much the light must bend in the optical path/design of the lens. The f4 having a larger optical design would need to bend and thus diffract the light more to achieve an aperture of f11 than a lens with an optical design with F5.6 as is maximum aperture.

Both will have to bend and deffract the light to focus the image on the sensor as light travels in straight lines. However when wide open, it will be at a level thats not as noticable as when you stop the lens down further and the light must diffract more for the smaller aperture opening.

Hope i'm not just jibbering.

I'm afraid you are!   

You're confusing two different processes - diffraction and refraction. Lenses use refraction, not diffraction, to "bend" the light and focus the image. But diffraction at the entrance pupil determines the resulting angular resolution and (in conjunction with focal ratio) the resulting spot size.

All "perfect" or ideal lenses, at the same f-stop, and the same wavelength(s) of light, give identical spot sizes. It doesn't matter whether they are f1.4 or f8 wide open; at f11 they will behave identically, and present identical PSFs and MTFs. It also doesn't matter whether one is a fisheye and the other a telescope. In linear, focal-plane units, they will perform identically on axis; although the angular resolution (on the subject) will of course be different if the focal lenghts are different.

So the original poster, Dave, was absolutely right to be incredulous of this sort of statement:
"One thing I noticed about the Rodenstock HR lenses, especially the wides is they say the ideal aperture is f5.6. Smaller apertures than that can cause diffusion issues. The Apo Sironar digital lenses have from f8-11 as ideal apertures. Even a 35mm lens at f5.6 doesn't have really great DOF, so it would seem that the 35mm Apo Sironar digital 35mm lens might be a more useful lens."
- this statement is not backed up by physics. For one thing, they seem to have confused diffraction with diffusion. For another, if by "ideal aperture" they mean the widest aperture at which diffraction-limited performance is reached, then the first lens to hit this performance is the HR one and it will surpass the other lens from f5.6 to f11 and then equal it from f11 down. That definitely renders the HR "a more useful lens".

There is just one nuance I would add, and it's that once real-world lenses are stopped down to one of their diffraction-limited stops, the shape of the aperture formed by the iris blades can be a factor in determining a "winner" - when I said that they would perform identically, that was for an ideal lens with a perfectly circular aperture. Non-circular shapes (evidenced by the angles where iris blades meet) lead to diffraction spikes; so the more circular the aperture, the less energy is thrown out from the centre of the PSF into such spikes, and the higher the MTF at low contrast.

Ray

[Dick Roadnight]

Lens choice is a complicated issue, especially for tech camera lenses, and the widest aperture, and the widest diffraction-limited aperture are two of many quantitative parameters.

The data in MTF charts is too complex to be expresses in a single number, but it indicates useful circle of confusion and decrease in res off axis, but you only need a large image circle diameter if you need it for shift, for e.g. for architecture or shift-and-stitch.

...and you could test the MTF for every tilt angle.