Repost: What makes a ‘good’ lens? (part I)

_1M01078 copy

Following an odd resurgence of emails lately about system matching, lens quality (on non-native systems), sample variation, decentering and similar topics – I thought it made sense to revisit this topic from the archives. ‘Which is the best lens for X?’ might seem like a very obvious question, but the moment you try to define a set of criteria to quantify ‘good’, you soon realize there’s quite a lot more to lens performance than immediately meets the eye. And this is before (but really should be much after) creative considerations, perspective etc. In any case: for those of you without the ability to try a large number of lenses – let alone samples of the same lens – how do you know if the one you’ve got is ‘good’?

When evaluating a new optic, I look broadly at several categories; this article will explain them a bit further. It’s wroth keeping in mind that most of these evaluations are relative and/or subjective; something that might perform well on a larger, lower-density sensor and balance properly on a large DSLR might not do so well on a very high density crop factor camera without a built in grip. It is therefore important not to consider ultimate resolution alone, but how well the system works together as a whole. A pancake zoom on a GM1 is a match made in heaven; you can mount an Otus on it, and it’ll deliver better resolving power, but it won’t be practical to use simply because of balance and manual focus/ stability/ ergonomic issues. Remember: the job of a lens is to collect and relay information, in the form of wavelength
(color) and spatial position to the recording medium. The less it affects the information along the way, the better.

Resolution
This is perhaps the most important and most easily defined category. Resolution at maximum aperture (and various points in the zoom range, if applicable) is obviously the priority; but one also has to consider whether the lens is symmetric: optical designs may intentionally be compromised (e.g. for long zoom ranges) to deliver high centre performance at the expense of corners or edges; however, they won’t intentionally be sharper on one side than the other. That would be an indication of manufacturing issues. The better the lens, the higher the resolution across the frame it will deliver at faster apertures. It isn’t difficult to make a f5.6 kit zoom with a moderate focal length range that delivers decent performance across the frame, but it’s quite something else if it delivers the same resolution at f1.4 and in the corners.

Correction for aberrations
In order for a lens to deliver high performance in the corners and at fast apertures, it must be well corrected: this is to prevent different wavelengths of light focusing at different points – both laterally and longitudinally. If the focal plane for different wavelengths shifts laterally, then you’ll see spectral separation in the form of lateral chromatic aberration, or blue-red/ green-red fringing. If you see it in front and behind your plane of focus, that’s longitudinal chromatic aberration. Lenses designed to minimize or eliminate this for visible wavelengths are usually given an Apochromatic or APO designation. Note that this doesn’t say anything about spreading/ ‘smearing’ of point light sources – that’s coma, and it’s possible to have an optical design that suffers from coma without having chromatic aberration (and vice versa). Aberrations that cause different wavelengths to focus at different spatial points for a given focus distance setting will land up reducing resolution because they affect the lens’ ability to resolve fine detail structures – i.e. edges – giving the overall impression of being a bit soft.

Microcontrast
This is a property of a lens that has to do with its ability to resolve the very finest detail structures – i.e. of low contrast and high spatial frequency – a lens with ‘good’ microcontrast is able to do this for structures that are close together/ fine and similar in luminance. Although you can increase the apparent gross contrast of a lens by sharpening, it’s almost impossible to ‘fix’ microcontrast because you cannot generate information that simply wasn’t recorded by the sensor to begin with.

Macrocontrast and flare
Overall contrast – recorded luminance between brightest and darkest areas of a scene that take up a significant portion of the overall image width – is quite dependent on the coatings of the individual elements; in order to maintain high contrast, you need to not have any stray light bouncing around inside the lens and landing up on different portions of the image than the area from which they came. A lens that has high flare has high internal reflections and poor coatings; this will affect both global (macro) and micro-contrast. It is especially important if you’re shooting into the light, as this will exacerbate the problem. The very best lenses show very little to no flare and have very good coatings.

Vignetting
I consider this to be the least of the optical maladies, since it’s easily corrected for in post processing – dark corners are not really a big deal so long as they aren’t completely black.

Transmission
If you’re losing light internally, or it’s being reflected out again at each lens-surface interface, then you’re not collecting it at the imaging plane. It is impossible to make a lens surface that has 100% perfect transmission, though it’s possible to minimize losses to the point that the vast majority of light makes it to your recording medium. Lenses with low flare and high contrast typically also have very high T (transmission) stops; the closer the T stop to the f stop (geometric aperture, or focal length divided by effective entrance pupil), the better the lens. Very good prime lenses have T stops that are usually within 0.1 of the F stop – f2/T2.1, for instance – whereas kit zooms that have an f stop of 5.6 may well have a T stop of 8 or lower. Not all f stops are equal – look at the shutter speed to give you clues.

Color
Transmission also affects color rendition: a lens needs to have equal transmission across all wavelengths of the spectrum in order to record an accurate reproduction of the scene; these are neutral. Lenses with color casts are attenuating portions of the spectrum, resulting in an overall shift in colors. Though the sensor or recording medium can affect this, you can generally see the difference between a very neutral lens and one that’s attenuated in a certain color. Neutrality is obviously desirable because it means you’ve got more information to work with later.

Focus shift
We change gears a bit here and look at the behaviour of the plane of focus as the lens is stopped down: if it moves backwards or forwards, then the lens exhibits focus shift. This means that you need to consciously adjust your focused distance in order to get the theoretical depth of field you expect; there can be significant differences between maximum aperture and say f4-5.6. Better lenses will employ floating elements to automatically correct for this as the focusing group is moved to the desired distance; this is computed as part of the lens’ optical formula.

Focus breathing
This is when the apparent magnification of a lens changes as you change focus distance; frequently, lenses will shorten their effective focal length in order to focus closer – this is so they require less helicoid extension and therefore can be made more compact. Although a consumer superzoom may reach its labelled maximum focal length at infinity, you will find that at minimum focus distance, your 18-300 might well be giving you magnification equivalent to 150mm. Let’s just say there’s a reason those 300mm supertelephotos require long barrels, but can only reach around 1.5m minimum focus distance: they do not shorten in focal length. The very best lenses do not exhibit focus breathing at all – cine lenses come to mind; they are designed this way to enable changes in focus plane (‘pulling focus’) to not change the composition of the scene. Few stills lenses are designed to compensate for this, partially due to size, partially because you can always recompose between shots – something that isn’t always possible in a video sequence. In fact, the only stills lens that immediately comes to mind that does not exhibit focus breathing is the Zeiss 1.4/55 Otus Distagon.

Field curvature
Ever have issues with your corners not being sharp, but something towards the front or rear of the frame being in focus instead? This is due to field curvature. If you see a resolution test where the sharpness (resolution) drops off alarmingly towards the edges of the frame, it’s probably because of this; it means that the plane of focus isn’t so much a flat, two-dimensional plane as a section of a much larger sphere. It isn’t always a bad thing because it can create a more three-dimensional effect to the image by making the edges effectively more out of focus than they should be. But it is worth noting for flat-field work – e.g. macro and reproduction.

Distortion
Straight lines should render as straight lines: if there are a lot of elements in the lens to correct for other aberrations across various focal lengths, then chances are they’re going to introduce some nonlinear projection. Some types of distortion – simple spherical pincushion or barrel – are easy to correct for; others like ‘wave’/ ‘moustache’/ or ‘sombrero’ types are trickier, and worse still, tend to vary with focal length. Of course, extreme distortion of any type – simple or not – is not desirable as correction may well result in a change in effective field of view.

Performance at distance
‘Macro’ lenses are designated as such not just because they have shorter minimum focus distances, but also because the optical design has to be optimized for near performance where the object and image distances are relatively similar. This requires (usually) extra glass and nonlinear helicoid movements at closer distances; which of course costs more. There’s a reason why a cheap 50/1.8 with an extension tube does not perform the same as a dedicated macro of the same focal length. By a similar token, not all lenses (including some macros) do well at infinity, either; faster lenses usually start to be compromised at longer subject distances. Very few lenses are good throughout the range, and most will have a ‘sweet spot’. It’s important to take this into consideration if you’re wondering why your fast prime does fine at f1.4 and 3m, but not f1.4 and infinity.

To be continued in part II

em>

__________________

Visit the Teaching Store to up your photographic game – including workshop videos, and the individual Email School of Photography. You can also support the site by purchasing from B&H and Amazon – thanks!

We are also on Facebook and there is a curated reader Flickr pool.

Images and content copyright Ming Thein | mingthein.com 2012 onwards unless otherwise stated. All rights reserved

Comments

  1. Metin Colak says:

    Dear Ming, good, informative article. However, I never understand why people takes ‘field curvature’ as a bad factor into consideration. A good controlled field curvature increases 3D in your image as you underlined. I think, modern lens designers have two option in lens design: Either they select sharpness, and optical perfection, or artistic aim. New lenses, including new Leica lenses, are more or less optically perfect. However they produce (with some exceptions of course), flat, 2D photographs. Everything seems perfect optically, clinically, however there is something lack a good artist, photographer percept. Compare for instance Leica Summilux-M 75mm f1.4 to Otus 55 f1.4. I know they are different systems. But I have both of them, and I can easily write here that one of them you can use more artistically, painterly: 75 Lux! Ok, Otus, or Otus like lenses, like 50mm ART, are optically very good, they produce clinically very sharp images, better colours, but flat and lack of that extensively used term, ‘character’. You can get different ‘blue’, ‘green’, black’ with 75 Lux, but most probably you will get very accurate reproduction of ‘blue’, ‘green’, ‘black’ with your new lenses which produce very sharp images, a reproduction of reality. To me, a photographer should select either s/he reproduces reality or reshapes it in photographic image.

    From this perspective, my old Zeiss Planar 85 f1.4 produces more artistic photographs than my new Milvus, or Otus line lenses. Similarly, no images produced by the Noctilux f0.95 can catch up the artistic level of the images produced by Noctilux f1, specifically at the widest aperture.

    Vignetting is also not bad, it add dimension to your images, if you are not taking landscapes.

    To me, modern lenses are very good especially in the field: Micro-contrast and flare.

    • Back to the original question, I think: ‘better’ depends on what criteria you are assessing the lens; not all images have to be flat-field razor sharp, but if I had a painterly macro lens with field curvature for product work, I’d probably throw it across the room 🙂 Similarly, there are a lot of less than perfect lenses in my arsenal (Hasselblad C T* 2.8/80, Hasselblad 100/2.2, CY Zeiss 2.8/85, Ricoh GR LTM 28/2.8 etc), but they have specific uses…

      • jean pierre (pete) guaron says:

        Surely this comes back to one of your other comments, Ming? A question of personal preference? And often this will depend on what, on the day, the photographer is trying to achieve?

        • Precisely – but it isn’t personal preference most people have the problem with; nor is it the answer they expect when they email me – it’s almost always to do with optics… 🙂

          • jean pierre (pete) guaron says:

            Another attack of GAS, perhaps. 🙂

            • Metin Colak says:

              It is not an attack of your so called ‘GAS’, it is a revolt to the comments, or analyses of so called ‘experts’ who put a lot of importance on resolution, and other controversial categories in the forums, or on websites which, in the end, reflect the problematic approach of the contemporary lens design, and photography.. Currently there is a tendency in photography aesthetics which I take ‘problematic’: The more detail you get the ‘better’ pictures you take! It is like a passion to see reality from a rather pornographic perspective..Maybe this is the effect of current resolution and flat view addicted capitalist cell phone culture… You never guess, or understand, or analyse the whole effect of this culture on you, your view, your eyes, and your final judgements.

              • jean pierre (pete) guaron says:

                Always wanting “the best ever lens” seems to me to be precisely an “attack of GAS” – especially when the end result with something like 99% of all the photos in the world today is that they are merely viewed on one form of video screen or another, with no theoretical possibility of exhibiting the clarity these lenses can produce. And the pursuit of theoretical “best” leaves the basics of producing a good photographic image lost, forgotten.

                • Metin Colak says:

                  Exactly, we cannot see the whole strength of a lens due to the limit of the screens, or medium on which we see them. I agree with you at this point, however I strongly believe that the capability of each lens can even be analysed, from an aesthetic, creative perspective at the same time on a limited medium…

              • Actually, I’d argue that the more is revealed, the more careful you have to be with your composition and observation…ultimately it only becomes problematic if you let it dominate over the creative elements.

                • Metin Colak says:

                  There is a paradox I hear in every interview of lens designer: When you increase detail in the final image using optical elements, glass technology, grinding methods etc you kill something which in turn become ‘creative’ in your final image. Either you should believe in such a high resolution exhibition, or appearance of reality, or re-shape it in creative ways with enough resolution. This is the dividing line of current ‘genius’ Leica lens designer, Peter Karbe, and old über master Walter Mandler. Where Karbe believes in transparency Mandler believes in ‘character’, and this is the division between secrecy and pornographic appearance of reality. The one (Karbe) lives under the effects of our current culture, the other (Mandler) lived and produced in the transition period, between the modern and postmodern.

                  My final words: There is an undeniable wish to see more and more, in infinite number of pixels of reality… Is this the cause of the crisis, or friction, between corporeal appearance of reality and creativity? We have more, more ‘perfect’ lenses, and cameras, in short equipments, on our hands, but we are less creative, less artistic. Actual reality, corporeal appearance of reality kills the artist.

                  • Martin Fritter says:

                    I am inclined to agree with you, although one risks coming off like a Luddite crank! I recently saw a gallery show of Burtynsky landscapes and came away sorely disappointed: they were just flat. Although horses for courses of course.

              • jean pierre (pete) guaron says:

                Spent a couple of hours yesterday seeing a classic example of what what I think you are referring to, Metin, when you refer to the ‘”flat view”. A film about the works of the artist Canaletto.

                I was struck by the difference between the “photographs” (stills) of his works and the filmed (cinephotography) versions. And then I realised – the photographers had gone to enormous lengths to produce a “perfect” picture – and in the process, they had destroyed something fundamental to Canaletto’s paintings. Vis – the “photographs” had obviously gone through some a rigorous process of lighting, post processing and printing to “bring up the detail” and to homogenise the contrast, that they looked rather flat, compared with the cinematographic versions. In the cine versions, the backgrounds were generally darker, the shadows almost black – but the “interest”, such as figures in the foreground, leapt out of the paintings. Even on a cinema screen, the cine versions showed paintings which seemed almost three dimensional. The contrast between the two versions was astounding.

                And I found I could not appreciate the “photograph” versions – because the flatness of the tonal range completely took away one of the two most striking features I was seeing in the paintings, in the cine versions – the vibrancy of contrast. (The other – his command of perspectives etc – is fairly indestructible, no matter how his paintings are reproduced).

                • My 2c on this: paintings are very much three dimensional and the medium itself interacts with light and vantage in different ways; a photograph can’t reproduce this because it’s a) two dimensional and b) both light and vantage are fixed. It is therefore up to the photographer to make a best guess at the intention/spirit/feeling of the painting in their executional choices…

  2. jean pierre (pete) guaron says:

    LOL – it seems I’m not the only person in the group who thinks your archive is a real treasure trove, Ming! I’m now looking forward to the updated version of “What makes a ‘good’ lens? (part II)” (PS – I shouldn’t be allowed NEAR the “fave” button, when I’m looking through a gallery of your photos! 🙂 )

    Thanks for giving me yet another reason for loving my Otus 1.4/55 – to be fair, I also have the Sigma ART 1.4/50, which I thought was the best I’d ever had (till I got the Otus), and I still use it for AF (the Otus, of course, is MF).

    Your concluding paragraph refers to issues of performance at longer distances (infinity etc).

    Lately, I’ve been exploring the possibility of acquiring a much longer focal length lens, and I’ve run into choppy waters. The fact that I can’t see myself buying a lens that costs $10 or $15 grand is my problem and of no possible interest to the group. But what are the other choices? I looked at zooms, and their manufacturers are very proud of their creations. But hang on – what is the point of a 200mm-500mm zoom, if the “sweet spot” is at 300mm and the image at 500mm is kind of soft focus/blurry? And I was thinking of buying a lens, not a vacuum cleaner – so why would I want something that sucks huge quantities of air into itself, from the outside, and fills both the back end of the lens as well as the camera body with fine particles of dust? (One reviewer drew attention to that issue, and said something along the lines “. . . over time . . . gradual build up . . . in the lens and the camera”. Not quite what I had in mind, when I set out to find something suitable to add to my arsenal). So back to looking around for a suitable prime – sigh!

    • Thanks – the only problem we have now is there’s so much material in the archive (1500+ posts) it just gets lost quickly…

      Longer lenses: it’s very difficult to separate performance at distance (where you’d likely use the much longer FLs) and atmospheric issues from lens issues; if you can’t get decent test conditions you may be in for surprises in either direction. Best you can do is test at mid distances and hope AF and resolving power etc. hold. The 200-500 VR is good, but enormous. The new 70-300 P VR is also good; I needed a bit more reach (but not size) so I went for the new Sigma 100-400. It seems all long zooms get a bit weaker at the long end because you’re basically magnifying the image circle; unless the initial resolving power is so high you can magnify without penalty (similarly, affecting TC use/performance) – but at that point, we’re talking silly money and sizes.

  3. excellentnas to be expected from you!

  4. Thank you for the informative article, Ming. Have you had the opportunity to evaluate the Nikon 28 1.4 E?

    • Not yet, unfortunately. I haven’t actually found a need for it in the way I shoot at the moment; that focal length and aperture combination tends to be more documentary-biased, for which I’d want something a little physically smaller than the D850 (and preferably with stabilization) – for the moment that’s been the Panasonic 15/1.7 on the PEN-F, or Leica 50/1.4 ASPH on the X1D.

  5. Interesting! I’m in the process of using my Mamiya 645 lenses with a digital body. Your article will give me the keys to find out if it worth trying.

    It’s funny too because yesterday I was reading an interview about how was created the last 50mm from Pentax designer and mechanical design team. It’s really incredible how they use physics and optical knowledge to design and build a lens AND to correct aberations induced by a chosen optical formula! A lens is not just aaperture and a focal distance. here is the link: http://www.ricoh-imaging.co.jp/english/products/star_lens/special/sp_dfa50-14/interview/

    Every manufacturer should do the same. It would be easier for the consumer to chose what lens to buy knowing what the designer wanted to achieve.

    Another thing they talk about is distortion is not always corrected. It depends of the focal lenght. Our brain will feel normal to see distortion with a ultra wide lens. Correctng every distortion in this case will give a picture not natural to our eyes. Our brain is part of the design 😉

    • I think Nikon posted something similar in the past. At the medium format end, we try to have a good balance of everything – a ‘clean’ rendering that allows the intention of the photographer to translate without influence from the lens. The tradeoff is of course maximum aperture, but given the size and weight (and cost) of very fast MF glass, we think this is reasonable most of the time…

  6. Thank you Ming! Do you recommend certain lenses or camera-lens combos that are best according to your findings? I had the Otus 55mm and I wish Zeiss gave us Mini-Oti for the mirrorless SoCaNikon. I think most (enthusiast) photographers could do with two lenses only and would like to pay good for those two beeing about perfect.

    • There are a lot of them…probably too many to list. ‘Best’ is also subjective: there are technically excellent combinations, and artistically excellent ones; they can be very different. Transparent lenses are brutal because they take no prisoners and you can’t hid behind the rendition; on the other hand, they don’t get in the way, either. Depends on your intention, I suppose!

      We (Lloyd Chambers and I) tried to persuade Zeiss to do a mini-Otus line at say f2.5-2.8, but no go – they were too busy doing mirrorless lenses sadly…

  7. Very Helpful

  8. great article! Thanks Ming

Trackbacks

  1. […] Panasonic GH5 vs Sony A7III Camera BATTLE ROYALE! (TheEverdayDad). What makes a ‘good’ lens? (Ming Thein). BEST Sony a7 III Gimbal Lens? (Jason […]

Thoughts? Leave a comment here and I'll get back to you.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: