The resolution of fine detail, lens quality, the selection of Leica lenses


There is much discussion (and not only in the Leica world) what characteristics in a lens are the most important for thhe creation of photographs with stunning impact, visually or by content. Some lenses are attributed with mythical properties, be in that ghastly old beast: "the Leica glow" or the modern counterpart: "bokeh". There is talk about three-dimensionality of the images, and some lenses are allegedly better in the reproduction of this visual phenomenon than others. Especially among collectors (and sellers of collectible items) there is a strong aversion against a scientific approach as exemplified by the MTF-graphs as an indication of optical performance.
Many Leica photographers, who advocate and practice the classical style of Leica photography, the snapshot, in their documentary, street or portrait photography, will shrug off any attempt to methodically describe the performance of a lens. This erroneous assumption is always at the bottomline of the debate: lens testing is done in a lab with flat or plane objects like testcharts or even a page of a daily paper and in real life there is no such thing as a plane object.
That last part of the argument is correct, but the first part is not. There is no compelling scientific argument that states that there can be no valid correlation between the optical quality of a lens as measured by a procedure that has a flat object as its target and its performance in the real world of objects with depth and width and height.
Representation of solid objects: shapes and shades.
Solid objects are, on analysis, defined by shapes of subject outlines and forms and shapes of details wthin these outlines. Subject outlines indicate the nature of the object(s) (ike a cat, a building, a car or a girl sipping a drink in a cafe). Shapes of details give further information about the subject, like the texture of clothing (the stripes or blockpattern in someone's jacket or the whiskers and tongue of a cat, licking tself.
In addition to shapes, any object is defined by tones and tonal differences or in other words differences in brightness values. The classical photographic picture has sharply dlineated outlines and is well graded in tone from deep shadows to highlights in fine incrments of tonal value. If I were to photograph a white vase with frontal lightning I get a picture that is flat and shows the outlines of a vase. To get depth into such a pcture, I need more subtle lightning effects, such that I can reproduce tonal differences (shadows and catch lights on the surface of the vase or even some hints of texture. Here we encounter the visual depth cues. Our eyes can interpret particular tonal differences or shades of grey as giving an impression of depth or plasticity. Go to an art museum and study the master painters of the 17th century to see how they did it.
A realistic representation of a solid object on a flat negative and/or print paper needs a fine balance of outlines of shapes and forms and in adittion a long scale of tonal differences. The arrangement of the shapes and the gradations within the tonal areas of small details bring the illusion of depth and thus plasticity.
Lens performance and image quality.
What we expect from a lens to represent our solid objects is clear: the outlines of shapes should be reproduced with a high edge contrast and for the definition of the fine tonal gradations or differences in brightness we need a lens that can capture even the smallest difference of tonality in an area. As example: the very finely differentiated shades of brightness in a face will convey the impression of plasticity. The lens should be capable of capturing these tonal differences without any loss. Our lens needs high contrast at the edges of shape delineations and high fidelity for the definition of small differences in tones. Both aspects (edge contrast and high fidelity) are measured with the MTF graphs and not all with resolution figures.
If we use a lens wide open, we have the usual problem of diminished capability of definition, compared to the optimum aperture. But we have also the problem that the depth of field is narrow. When we are outside the sharpness area of the lens, what we 'see' as unsharpness blur is in fact image degradation, which is defined by the aberration correction of the lens.
Older lens designs are not as well corrected as current ones. Both the sharpness area, as well as the unsharpness areas (backgrounds and foregrounds) exhibit more aberrations than modern lenses. As we do not percieve differences in unsharpness as well as we can detect faults in the sharpness plane, the difference between the quality of the sharpness and unsharpness areas in the print are less pronounced. We tend to interpret such a picture as having a more extended depth and offering a more realistic representation of what we assume is "really out there".
If we use older lenses at the wider apertures, they tend to become overall softer and the primary aberrations are more visible in halo effects, flare and mushy details at the limit of resolution. For the eye this may be quite pleasing as the perception of high contrasts is heavy on the eye and so there is a natural tendency to evade the confrontation of this visual contrast.
We should however draw a distinction between a pleasing picture and an accurate picture. Accuracy of representation will give a better and more lifelike drawing of the solid oject and by definition its depth impression. A more pleasing represenation, even when less accurate, may be more forgiving of the faults in the imaging chain and so give the eye some additional space for subjective interpretation.
What do we need?
Often we are hearing or reading that we should select lenses or film emulsions based on resolution figures. The idea is simple. The finer patterns of lines in a resolution chart represent the fine details in an object and when we can capture the lines, we automatically capture the details. This flow of reasoning is correct, but what value should we aspiring for. Older lenses are often credited with superior resolution figures of 300 lines per mm and this seems to be the minimum. A lens that has 'only' 60 lines/mm is crap, so the argument goes.
(I will not make a distinction here between lines per mm and linepairs per mm, as it is irrelevant for the argument).
Below is a picture of a part of a test chart. In the original the number of '1' represents a true physical space of one line pair per mm.



The number '4' represents 4 line pairs per mm. If I take a picture at a distance of 100 times the focal length, and on the film the pattern of '4' would be visible I have a resolution of 400 lp/mm. If I see the '1', I have a resolution of 100 lp/mm.
Extensive testing by me (see next article) has shown that the best film/lens'developer combinations can reproduce a pattern around 100 to 120 lp/mm on the negative. We need then a stable tripod, careful focussing and slow speed film, like Technical Pan.
If we relax are taking conditions, but still use slow speed film like Delta 100 or Tmax 100, we may be very happy if 60 (or at best 80) lp/mm would be recordable. Or we shift from tripod to flash. Flash will reduce the vibration of the camera, as the flash speed of 1/500 to 1/10.000 sec. will freeze all movement. In addition the sharp flashlight will enhance the contrast of the fine textural details.
If we change to handheld photography, we may be pleased wen we see 30 to 40 lp/mm at the higher speeds (above 1/500 at least).
When we use grainier ISO400 material, slower shutter speeds and bad conditions, (fast focussing, moving targets etc), we should be surprised to get 20 to at most 30 lp/mm on the negative.
That is a far cry from the original high figures and with these results a good quality inkjet printer can handle this resolution too. Remember that a 20x25cm print has an enlargement factor of 8, and the 20 lp/mm on the negative are reduced to 2.5 to 3 linepairs/mm in the print.
The comparison of lenses and the selection of what features are important, can not be discussed as is done in most popular articles or discussion groups: that is without looking at the imaging chain and the results that are demanded and the situations where the photographs are taken.
The realistic value of 20 to 30 lp/mm is also on a level that most lenses (older and curent Leica lenses, but also many lenses of other manufacturers: like Voigtlander, Zeiss, Canon and Pentax and many others) can capture without much problems.
If we restrict ourselves to this level of image quality and lens comparison, based on these parameters, it is quite inderstandable that many results at first sight look equal and that it seems hardly possible to differentiate between whatever lens we use. The many wars that rage between photographers about which lens is best and can I see a difference between a picture made with a Pentax lens or a Leica lens, are in fact the result of only using a fraction of a lens' potentail or loking at the wrong clues for image quality
When we restrict ourselves to simple 'tools' like resolution figures or the visual inspection of pictures at a low magnification (below 8 times enlargement), the differences between lenses may be small to the casual eye. The discerning crafsman however will clearly detect the more important aspects of image quality, as the distribution of image quality over the picture from corner to corner, the crispness of textural details, the smootness of gradation in very small picture areas, the distribution of light in the specular highlights etc.
Two lenses that have the same figures of (measured) resolution, can differ significantly in the way this resolution is represesented. The drop from 10 to 60 lp/mm may be quite steep in one lens and very gradual in another. Contrast over the whole spatial ramge will drop too, but again the rate of drop is an important parameter, which will be visible in the sparkle and cleanliness of the definition of the lens in question.
Careful comparison tests and a careful selection of subjects is necessary to make meaningful statements in this respect. This however is lacking in most discussions about lens evaluation, objective and subjective.
And then the discussion will become self repeating and without any clear or definite conclusion.