On Thu, 8 May 2003 23:19:54 +0100, David Littlewood <david@nospam.demon.co.uk> wrote:

>In article <5779d5d6.0305081215.2df1d183@posting.google.com>, Chris

>Wilkins <chris.wilkins@virgin.net> writes

 

>>Are lenses better sharper at one aperture setting compared to another?

[...]

 

>"Sharpness" is a slightly imprecise term. There is:

>

>(1) Resolution - the ability to distinguish two very close objects

>(points, lines etc.).

>

>(2) Contrast - the differentiation between areas of different tone.

>

>(3) Acutance - the sharpness of the boundary between areas of different

>tonality.

>

>Acutance is largely a function of the film and the developer used to

>process it; the lens has little effect on this. A negative or

>transparency of very high acutance may look sharper than one of low

>acutance even if the latter has a higher resolution, until you look very

>closely through a magnifier or a microscope. Since it is not much

>affected by the lens, I assume this is not what you are interested in.

>

>Contrast is affected by both lens and film/processing. Again, a high

>contrast result may look superficially sharper than one of low contrast

>which has higher resolution.

>

>Resolution is the parameter which people normally refer to when talking

>of "sharpness", even though the other two may actually be more important

>in many kinds of image.

>

>Resolution is affected by both lens and film/processing. It is usually

>measured in lp/mm - line pairs per mm (the finest pattern of black and

>white lines which can be distinguished on the film). Just looking at the

>effect of the lens, this is a balance of two factors:

>

>First, the resolution of the lens can never exceed the diffraction

>limit. This is determined entirely by the effective lens aperture and

>nothing else (well, actually, by the wavelength of the light as well -

>but I assume we are talking about white light here). The resolution is

>approximately 1600/effective f-number of the lens in lp/mm. ("Effective"

>lens aperture means that if the lens is being used for macro work the

>reduction in aperture must be taken into account - for instance, at 1:1,

>the effective aperture of f/8 becomes f/16). [Note some people use

>1500/f-number or 2000/f-number - there are different views about what

>can be "seen" to be resolved.]

>

>For example, a lens at f/8 has a diffraction limit of 1600/8 = 200

>lp/mm. Used in macro at 1:1 it would be 1600/16 = 100 lp/mm. Used wide

>open at say f/2 it would be 1600/2 = 800 (but of course the film would

>not cope - see below).

>

>If this were the only story, all lenses would be best wide open.

>However, real lenses have aberrations - deviations from theoretically

>perfect behaviour. There are several of these, and most of the ones

>which reduce resolution (spherical aberration, coma, astigmatism, and

>longitudinal colour) are reduced by using a smaller lens aperture (i.e.

>a higher f-number). Only lateral chromatic aberration reduces resolution

>and is not reduced by stopping down, but it only causes serious problems

>in very long large-aperture lenses.

>

>So, a perfect lens would give the best resolution wide open. A real lens

>for normal photography will have its aberrations reduced by stopping

>down, and in most cases the improvement is, up to some intermediate

>aperture, more than the deterioration from diffraction. A really bad

>lens may need to be stopped down to f/16 before the aberrations are

>controlled and the performance deteriorates from diffraction. A really

>good lens will typically work best at about 2 stops down from maximum,

>maybe 3. Thus f/5.6 or f/8 may well be the aperture giving the highest

>resolution in many cases for very good 35mm lenses, f/11 for poorer

>ones. This accounts for the resolution chart graphs you sometimes see in

>magazines, with performance rising to a peak somewhere in the middle.

>

>There are lenses for which performance is best wide open; however, these

>are usually limited in some other aspect of their design, such as having

>a very narrow field of view (many scanning lenses) or to work only at

>one set of conjugate distances (most microscope lenses). The very best

>enlarger lenses give best results 1-1.5 stops down.

>

>Of course, as suggested above, lens resolution is only half the story;

>film resolution (plus the resolution of whatever you use to "see" the

>film - loupe, scanner, enlarger, projector or whatever). The interaction

>between the two is complex, but the overall result will always be worse

>than any of the component parts.

>

>...And this glosses over the interaction between resolution and

>contrast, and modulation transfer functions. If you want to go into this

>further I suggest you get hold of a book on optical theory.

>

>I don't know the answer to your questions about particular lenses, but I

>would guess about f/8-f/11.

 

Great answer!