Why are lenses unsharp wide open?

Discussion in 'Digital Cameras' started by Alfred Molon, Jul 30, 2011.

  1. Alfred Molon

    Alfred Molon Guest

    Just wondering - does it come from physics or is it pure economics, i.e.
    too expensive to make a lens with corner to corner sharpness at F1.4?
     
    Alfred Molon, Jul 30, 2011
    #1
    1. Advertisements

  2. Spherical aberrations if nothing else.

    The smaller the spherical surface area that is ground, the closer it
    comes to focusing all points of light correctly. The larger it is
    the hard it is to get the curve over the entire area exactly the same.

    The opposite is true for diffraction though, as the larger the aperture
    the smaller the percentage of light rays that are affected by the edges
    of the aperture. At a very small aperture virtually all of the light
    going through the tiny hole will suffer some diffraction.

    Hence at some point between minimum and maximum aperture, the lens
    is sharpest.
     
    Floyd L. Davidson, Jul 30, 2011
    #2
    1. Advertisements

  3. Alfred Molon

    Ray Fischer Guest

    Yep. People don't want to pay the price.

    Do you?
     
    Ray Fischer, Jul 30, 2011
    #3
  4. Alfred Molon

    Robert Coe Guest

    : >Just wondering - does it come from physics or is it pure economics, i.e.
    : >too expensive to make a lens with corner to corner sharpness at F1.4?
    :
    : Spherical aberrations if nothing else.
    :
    : The smaller the spherical surface area that is ground, the closer it
    : comes to focusing all points of light correctly. The larger it is
    : the hard it is to get the curve over the entire area exactly the same.

    My understanding is that spherical aberration is not due to inaccuracies in
    the grinding of the lens surface; it's an unavoidable consequence of the index
    of refraction of the glass and the angle of incidence of light rays striking
    near the edge of the lens surface. A perfect spherical surface would still
    have it. That's why they make aspherical lenses, whose radius of curvature at
    any point on the lens surface varies with its distance from the center line of
    the lens. Aspherical lenses are more expensive to manufacture, but they can be
    made to correct for most of the effect of spherical aberration.

    : The opposite is true for diffraction though, as the larger the aperture
    : the smaller the percentage of light rays that are affected by the edges
    : of the aperture. At a very small aperture virtually all of the light
    : going through the tiny hole will suffer some diffraction.
    :
    : Hence at some point between minimum and maximum aperture, the lens
    : is sharpest.

    Without the word "hence", that statement would be true. Lenses are typically
    sharpest near the mid-point of their aperture range. But the implication that
    it's simply because it splits the difference between the maximum effects of
    diffraction and spherical aberration is an oversimplification. In fact, lenses
    are designed to have that characteristic, so that, on average, *any*
    undesirable optical effect won't be too pronounced at the extremes of the
    range. For example, my wife and I had two 50mm lenses for our Nikon film SLRs:
    an f/2 and an f/1.4. (Both topped out at f/16.) You might think that the only
    difference between them was the extra stop at the low end; but in fact the
    f/2's point of maximum sharpness was, by design, one stop higher than that of
    the f/1.4. (I believe those were f/5.6 and f/4, respectively.) So while the
    f/1.4 was better in low light, the f/2 was a slightly sharper landscape lens.

    Bob
     
    Robert Coe, Jul 30, 2011
    #4
  5. You are correct. I didn't say that quite right, as
    instead of "to get the curve over the entire area ..."
    what I should have said was to get the _refraction_ to be
    the same over the entire area is not possible with a
    spherical surface that is large. Hence grinding the
    correct curve is difficult, simply because it is
    aspherical.
    I think that is precisely correct as stated, and that is
    often just about that simple.
    The f/16 aperture pretty much means that if all else is
    equal the upper end of the range will suffer diffraction
    equally. I dont' see what is so odd about an f/1.4 lens
    being best at f/4 and an f/2 lens at f/5.6? In each
    case the trick is stopping down two stops to avoid
    spherical aberrations at the edges of the lens surface.

    You are quite correct about intentional design
    characteristics though, and certainly 50mm lenses are a
    great example. Virtually everyone makes a 50mm f/1.4
    lens of significant quality, and virtually everyone has
    at some time made an inexpensive 50mm f/1.8 or f/2 lens.
    Every 50mm lens designed to be very inexpensive has used
    the simple trick of overcorrecting for spherical
    aberrations to make just slightly out of focus areas
    look sharper than they would otherwise. The tradeoff is
    that at more than just slightly out of focus the effect
    is very harsh bokeh. The Canon "nifty fifty" and the
    Nikon 50mm f/1.8 AF-D lenses are classic examples.
     
    Floyd L. Davidson, Jul 31, 2011
    #5
  6. Alfred Molon

    Rich Guest

    Because lenses are FILLED with ancient, outmoded spherical optics that are
    cheap to make but have middlin results wide open.
     
    Rich, Jul 31, 2011
    #6
  7. Alfred Molon

    Alfred Molon Guest

    But... would it then be possible to have a "perfect" lens, sharp wide
    open, if the aspherical elements were made very precisely? Is it then
    just a matter of cost?
     
    Alfred Molon, Jul 31, 2011
    #7
  8. But... would it then be possible to have a "perfect" lens, sharp wide
    Yes, Alfred, but it's likely that if the lens was perfect for spherical
    aberration (given enough precisely-shaped elements), there would be some
    other issue which would make it less than "perfect" in some other
    respect - chromatic aberration, linear distortion, flare, bokeh,
    transmission loss, etc. etc. etc.

    Cost is certainly a major issue, so what you buy is a compromise between
    what you want and what you can afford! There would be little market for a
    very expensive lens with only marginally better performance.

    Cheers,
    David
     
    David J Taylor, Jul 31, 2011
    #8
  9. Alfred Molon

    Martin Brown Guest

    There is nothing in the physics stopping you making a fast lens or more
    easily a catadioptric mirror lens with any arbitrary focal ratio if cost
    is no object. However, the costs get extremely high for fast optics and
    the circular field of view that is truly diffraction limited is not all
    that large.

    But it will cost a lot and to get the absolute optimum focal plane
    diffraction limited image in something that can be realistically
    manufactured you often have to live with a curved focal plane.

    The problem is that for slow focal ratio lenses f8 and higher the small
    angle approximation sin(x) ~= x approximation is a good one. Faster
    lenses have to correct for ever higher terms and something has to give.

    What kills it is that the surfaces become incredibly hard to make and
    the position of every element absolutely critical. So you can design
    something that will raytrace OK in theory but is unmanufacturable!

    The mirror based telescopes suffer only from the geometric aberrations -
    any lens based system has chromatic aberration to consider as well.
    Digital cameras that can now tolerate lateral chromatic aberration and
    correct for it in software post processing have given the lens designers
    an extra degree of freedom to play with.

    A paper describing a state of the art fast survey telescopes design
    which are limited only by geometrical aberrations is online at:

    http://adsabs.harvard.edu/abs/1984MNRAS.210..597W

    And a much more recent review of f1 optics that are being designed for
    finding Near Earth Objects in a survey instrument.

    http://www.amostech.com/TechnicalPapers/2007/Telescopes_Instrumentation/Ackermann.pdf

    These are *very* expensive designs to implement.

    Regards,
    Martin Brown
     
    Martin Brown, Jul 31, 2011
    #9
  10. Alfred Molon

    Allen Guest

    A very good post, Martin!
    Allen
     
    Allen, Jul 31, 2011
    #10
  11. Alfred Molon

    Bruce Guest


    Economics has a big part to play. Many Leica lenses M lenses are
    superbly sharp wide open, but they cost a lot more than more prosaic
    Canon and Nikon glass that performs best when stopped down.

    My current favourite lens is the Leica 24mm f/2.8 Elmarit-M. It is
    sharpest in the centre at f/2.8, and across the whole frame at f/4.
    There is no benefit to sharpness from stopping down any further.
     
    Bruce, Jul 31, 2011
    #11
  12. Not all lenses are unsharp wide open.

    Both Canon and Nikon make tele lenses (300mm and up) that
    are sharpest wide open. The field of view of a 300m lens
    is so narrow that the off axis aberrations are not
    a terrible problem. The main problem is chromatic aberration
    and spherochromatism. Spherical aberration (at one wavelength)
    is no longer a problem at all, due to replicated aspheric optics.
    Of course, if they are sharpest wide open in means 1/4 wave
    accuracy summed over all elements ... and for say a 300mm
    f/2.8 lens this is not easy!

    Look at the MTF curves for Canon's 400mm f/2.8L IS II USM,
    they are superb. But also $11,000. Or the 500mm f/4L IS II USM.




    But the f/1.4 50mm lens (on a full frame 35mm camera) is
    a terrible design problem. First, you have to stuff all that
    light through the standard lens mount. Except for Canon
    this is a really big problem, and even they are not immune.
    Of course, some designers intentionally limit off-axis
    illumination to avoid the difficulties of correcting
    at f/1.4 across the field.

    With aspheric optics of course nowadays it is taken for granted
    that at one wavelength on axis there will be negligible spherical
    aberration. But the off axis aberrations are exceedingly difficult
    to control, even with multiple aspheric surfaces, as is spherochromatism.
    While the long teles can get most of their chromatic correction
    across the whole spectrum OK by putting most of the positive
    and negative powers in glass types that "naturally" match
    each other and cancel, that is impossible in fast lenses, since
    the shape necessary to control astigmatism makes using such (low index)
    glasses as the main power elements infeasible. So the chromatic
    errors of several elements have to have "delicate" cancellations.


    Doug McDonald
     
    Doug McDonald, Jul 31, 2011
    #12
  13. Alfred Molon

    Rich Guest

     
    Rich, Aug 3, 2011
    #13
  14. Alfred Molon

    Rich Guest

    That's enviable performance. One of the shocks you get going from 4/3 to
    APS or FF is the terrible edge performance of all but the best lenses,
    wide open anyhow. Using a Zeiss 35mm f2.0 against Nikon's very good but
    inexpensive 35mm f1.8 was a shock, with the Zeiss being much sharper wide
    open, and I wouldn't have described the Nikon as bad wide open by any
    means.
     
    Rich, Aug 3, 2011
    #14
  15. Alfred Molon

    Bruce Guest


    Enviable indeed. I bought my Leica 24mm f/2.8 Elmarit-M ASPH (I left
    out the ASPH in my previous posting) at a time when Leica prices were
    much lower than they are now. I probably couldn't justify paying
    today's prices.

    At around the same time I bought a Leica 35mm f/1.4 Summilux-M ASPH,
    another outstanding performer. Both lenses are now worth, in Sterling
    terms, somewhere between three and four times what I paid for them. Of
    course Sterling has dropped in value during that time thanks to the
    UK's indebtedness. Even so, these lenses have been a remarkable
    investment. I'm tempted to cash in and buy a Leica M9 body with the
    proceeds, but then I would need some lenses ...


    The Carl Zeiss Distagon 35mm f/2 is a beautiful lens. As long as you
    don't mind focusing manually, it is a far better choice than any 35mm
    lens Nikon has ever made. And if you really need f/1.4, there is the
    outstanding Samyang which is optically marginally superior to the
    spectacularly expensive AF-S Nikkor.

    Of course all these superb lenses are completely wasted on the vast
    majority of buyers who want zoom lenses = preferably with a large zoom
    range so they don't ever have to take them off their cameras - and are
    prepared to accept lousy optical performance in return.

    Pearls before swine ...
     
    Bruce, Aug 3, 2011
    #15
  16. Alfred Molon

    George Kerby Guest

    Forget something, Rich, you senile old bastard?
     
    George Kerby, Aug 3, 2011
    #16
  17. Alfred Molon

    PeterN Guest

    Why complain. He used fewer electrons to say hat he usually says.
     
    PeterN, Aug 3, 2011
    #17
  18. Alfred Molon

    George Kerby Guest

    True that.
     
    George Kerby, Aug 3, 2011
    #18
  19. Alfred Molon

    Robert Coe Guest

    : >: >>>
    : >>>Just wondering - does it come from physics or is it pure economics,
    : >i.e.
    : >>>too expensive to make a lens with corner to corner sharpness at F1.4?
    : >>
    : >>
    : >> Economics has a big part to play. Many Leica lenses M lenses are
    : >> superbly sharp wide open, but they cost a lot more than more prosaic
    : >> Canon and Nikon glass that performs best when stopped down.
    : >>
    : >> My current favourite lens is the Leica 24mm f/2.8 Elmarit-M. It is
    : >> sharpest in the centre at f/2.8, and across the whole frame at f/4.
    : >> There is no benefit to sharpness from stopping down any further.
    : >>
    : >>
    : >
    : >That's enviable performance.
    :
    :
    : Enviable indeed. I bought my Leica 24mm f/2.8 Elmarit-M ASPH (I left
    : out the ASPH in my previous posting) at a time when Leica prices were
    : much lower than they are now. I probably couldn't justify paying
    : today's prices.
    :
    : At around the same time I bought a Leica 35mm f/1.4 Summilux-M ASPH,
    : another outstanding performer. Both lenses are now worth, in Sterling
    : terms, somewhere between three and four times what I paid for them. Of
    : course Sterling has dropped in value during that time thanks to the
    : UK's indebtedness. Even so, these lenses have been a remarkable
    : investment. I'm tempted to cash in and buy a Leica M9 body with the
    : proceeds, but then I would need some lenses ...
    :
    :
    : >One of the shocks you get going from 4/3 to
    : >APS or FF is the terrible edge performance of all but the best lenses,
    : >wide open anyhow. Using a Zeiss 35mm f2.0 against Nikon's very good but
    : >inexpensive 35mm f1.8 was a shock, with the Zeiss being much sharper wide
    : >open, and I wouldn't have described the Nikon as bad wide open by any
    : >means.
    :
    :
    : The Carl Zeiss Distagon 35mm f/2 is a beautiful lens. As long as you
    : don't mind focusing manually, it is a far better choice than any 35mm
    : lens Nikon has ever made. And if you really need f/1.4, there is the
    : outstanding Samyang which is optically marginally superior to the
    : spectacularly expensive AF-S Nikkor.
    :
    : Of course all these superb lenses are completely wasted on the vast
    : majority of buyers who want zoom lenses = preferably with a large zoom
    : range so they don't ever have to take them off their cameras - and are
    : prepared to accept lousy optical performance in return.
    :
    : Pearls before swine ...

    But aren't the optical advantages of even a fine prime lens largely wasted if
    you end up having to crop the image to a quarter of its original size because
    you couldn't zoom?

    Of course the standard reply to that question is, "Well, you should zoom with
    your feet." But if there's a river, or a busy highway, or two or three
    referees, or 100 meters of vertical separation in your way, ...

    Bob
     
    Robert Coe, Aug 4, 2011
    #19
  20. Alfred Molon

    Bruce Guest


    So that's why so very few good pictures were taken before the
    introduction of zoom lenses. ;-)
     
    Bruce, Aug 4, 2011
    #20
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.