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


This 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. So, 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’?

Continued from part I.

Today, we start with what is probably the most subjective of all of the pictorial qualities of a lens. Generally speaking, ‘good’ bokeh isn’t distracting. Distractions are caused by areas of an image that unintentionally stand out; in the case of bokeh it takes the form of double images, hard edges to highlights, geometric shapes of highlights, and color fringing. Few lenses are optimized to also deliver neutrality in the out of focus ares – this requires correction for spherochromatism and longitudinal chromatic aberration. The very best lenses have a smooth transition from in-focus to out-of-focus areas with no odd artefacts in the intermediate zone; subjects should be recognizable but not distracting.

Ergonomics, build and ease of use
I think this category is pretty much common sense: are the rings located easily to hand? Do the switches move accidentally? Is it robust, weather-sealed and does the hood stay in place? Do the extensions wobble if it’s a zoom? Is it front-heavy? Bottom line: do you feel comfortable using it, or do you have to consciously change the way you shoot to accommodate the lens? A good example is the Nikon AFS 200/2 VR: optically, it’s one of the greats. Ergonomically, it always throws me off balance: it’s not very large, but very front-heavy, and the hood is far wider than it needs to be – so it can reverse over the stubby tripod foot that usually creates mounting issues with bodies that have a vertical grip because there’s no longer sufficient clearance between that and a large tripod head.

Autofocus performance – speed, accuracy and ability to precisely accommodate small changes in subject distance – is a function of both the lens and the camera’s AF system. Most of this depends on the type of motor used to move the focusing group; ring-type coreless DC motors (USM, HSM, SWM etc) can move quickly and in small increments. Lenses which are body-driven will have some backlash in the gear train that can cause slight focus ‘misses’ that become an issue with small changes in subject distance and fast apertures. In such situations, it’s often better to completely defocus the lens and start again to eliminate slack in the gears/ helicoid.

Manual focus and spacing/ gearing
Here’s another highly subjective quality. Some people prefer tight rings, others loose; regardless, backlash/ slop is not desirable because it makes small changes in focus distance difficult. This is usually the result of a helicoid with tracks that are slightly loose, or multiple helicoids not meshing properly. You also don’t want a lens that doesn’t have sufficient throw across the distance range – it’s difficult to hit a precise difference if only a tiny movement is required to go from infinity to 2m, and fast gearing also tends to increase the impact of backlash since tolerances remain similar, but will have a larger absolute effect.

Lenses that have increasing reliance on electronic components – motors, stabilisation actuators, focus clutches, distance confirmation chips, buttons, etc – both have a greater potential number of failure points, as well as limiting compatibility. Put it this way: you can mechanically adapt a 50-year old lens to a modern mirrorless body, but modern DSLR lenses may not even have aperture control. Though these protocols can be reverse-engineered, they don’t always work very well…

System matching/ performance
Optical formulae are always derived with an ideal back focus distance in mind. Changing this will inevitably change the optical performance of the lens. Beyond this, digital photography has introduced a new complication: the sensor surface is not flat and relatively homogenous at the microscopic level, but rather takes the form of a collection of pits – each of which contains a photosite. This affects both light collection efficiency and introduces potential optical issues due to shading and interference effects with the walls of the pits. Though modern sensor designs now incorporate microlens arrays to increase light collection efficiency – effectively focusing light from the sensor directly onto the center of the photosite – they also add another optical element into the system. Perhaps the most obvious example of this interaction is when you look at the corners of older wide angle lens designs; the output rays are not telecentric – i.e. they subtend from the exit pupil of the lens at an angle – which creates not only vignetting/ shading at the photosite level, but potentially also causes added chromatic aberration and color shifts due to interaction with the microlenses. In effect, each microlens for each individual photosite is another optical relay within the system: it too has to obey the same laws of physics as a complete lens. Although offset microlenses can compensate for this to some degree, modern ‘designed for digital’ lenses almost always perform better as they tend to be telecentric to begin with – thus negating any potential radial effects.

Ergonomic and electronic issues aside, it is also for this reason that dedicated lenses tend to perform better than adapted ones – aside from potential planarity issues leading to decentering and tilting. There are of course exceptions to both rules, but in my experience with dozens of different sensors and hundreds of lenses, this is generally the case.

Caution on using converters, filters and adaptors
Again, though there are always exceptions to this rule, additional optical elements will degrade overall system performance because lenses were not designed for to accommodate them in the first place. It’s why using a teleconverter on a fast wide will produce terrible results compared to a normal prime, and why superteles tend to not exhibit very much degradation with teleconverters. It’s also why cheap filters are a bad idea: the glass may not be perfectly planar, the coatings may be poor and introduce flare/ reduce transmission, and if they break or crack – you’re still going to land up scratching the front element anyway. As for adaptors between systems/ mounts – just bear in mind that you are introducing an additional pair of mount surfaces into the equation, both of which may potentially decenter or tilt the entire optical assembly. It is even more apparent with a high resolution sensor; scientific tests by various other sites such as Lensrentals have found this to be the case, and it matches with my own experience. Never mind the fact that you’re going to at very least lose AF and gain size/ bulk…

A note on built in lenses
Built-in lenses take two general forms: zooms in utility (‘Swiss army knife’) cameras and high quality primes (GR, Coolpix A, X2 etc). There are also some good zooms – the RX10 and X Vario come to mind. The higher end of this spectrum tends to perform very well indeed; the GR’s lens has a bite to it that is matched by few – if any – interchangeable lens solutions of a similar size. This is simply because these lenses are almost always designed to match the sensor; the entire optical system is optimized from the start. Practically, this means a GR may well be a better solution than looking for a good 28mm-equivalent for your APSC or 24MP FX DSLR.

Lastly: there’s always that special ‘something’ – an indefinable quality that makes you simply like the quality or rendering style of the output. Quantitatively, it’s a combination of many of these factors; however, it can be harder to put a finger on. Although a good lens may have high resolution, flat field, no distortion, uniform spectral transmission and low chromatic aberration, you might also like one that has moderate resolution, strong field curvature and warm transmission because it renders skin and portraits beautifully. I find that lenses with a very distinctive ‘signature’ tend to be quite polarizing: either you like the look, or you don’t; if you don’t, there’s not much point in buying one since you can’t do a lot to neutralize it. However, the opposite is also true: a lens that’s completely neutral and ‘transparent’ adds no character of its own: it’s all down to the photographer. Transparent or not, it’s still down to the photographer. MT



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 | 2012 onwards unless otherwise stated. All rights reserved


  1. Calvin Yee says:

    Hi Ming, Hasselblad just announced the XV adaptor, how would the Zeiss lens do with the adaptor? I can’t help noticing the cfe if 40.Thanks. Calvin.

    • They perform pretty much identically to the CFV-50C/ V series – sensor and total optical distance are identical. But honestly, unless a particular lens has a specific focal length or rendering quality, the native lenses are a lot more convenient in size (and the shutters are fully functional to 1/2000s).

  2. A very comprehensive and useful compilation of attributes Sir.

    A few things that I sorely miss on most new generation lenses are lack of aperture control rings, IR focusing scales and mount for rear filters.

    • It’s a mix of cost cutting and electronics; to put manual interfaces in for aperture etc. requires a lot of additional parts and sealing to maintain weather resistance. Rear filter mounts are problematic because you’re placing additional (unplanned and variable depending on the user) optical components in a sensitive portion of the optical path, and given digital’s sensitivity to any deviations from telecentricity this has understandably been avoided. More concerning though is that most of today’s lenses are going to be bricks if any of the electronics fail once there are no more spare components…

      (Unrelated: congratulations on comment #90,000 by the way…!)

      • Precisely. Once they run out of spares (which will likely be sooner than later), the lenses pretty much become useless. I do hope going forward manufacturers at least try avoiding ‘focus by wire’ mechanism and provide manual override option in their lenses.

        As you have rightly indicated, rear filters increase the probability of optical imperfections. But they also have quite a few benefits. Cost for one as they are smaller and cheaper. They also tend to stay cleaner and this is especially useful for gel filters which are prone to debris ingress. Thirdly they make you less inconspicuous (this can be a problem if you need to use IR filters in areas under high surveillance).

        Glad I was the 90K guy 🙂

        • Actually, the biggest benefit of rear filters is for lenses that have front elements that might otherwise be un-filterable; unfortunately these super-domed optical designs tend to also be the kind that are most sensitive to extra things in the optical path. A drop in with a normal ‘blank’ is probably the best overall compromise.

          Focus by wire is a necessary evil for CDAF: the lens has to be able to overshoot infinity to be able to determine maximum contrast at infinity, and you have to have some slip in the focus clutch. And unfortunately this of course means you can’t have hard infinity stops; the easiest way to get around this is focus by wire since you no longer have to physically link the focusing group drive and external physical controls, allowing for smaller and simpler lenses. But the feel is almost always rubbish, with just enough backlash to cause precision focusing issues…or so much slowness as to be frustrating.

  3. Martin Fritter says:

    Mose “lens reviews” are useless. The industry is converging and everything above a certain point tends to look the same. The examples used to illustrate the lens are always almost always just awful photographs. I can think of few counterexamples. The following is a link to the single best lens review I’ve ever read.

    • Martin Fritter says:


    • I’m flattered, but basically have one principle for these things: does it let me make the images I want, or are there glaring limitations? Or something else noteworthy in its physical properties? Also, something I’ve been saying for a long time: reviewers aren’t photographers, and the conditions often are not representative of actual use, therefore rendering the opinion of limited use… 🙂

      There’s only one problem with my review: the lens isn’t available! 😛

      • Martin Fritter says:

        I wasn’t just pandering – well not totally. You are an excellent photographer, you took excellent pictures using the lens and your enthusiasm was obvious. So here’s a lens that doesn’t check a lot of boxes and is kind of odd – which seems to be a Voightlander thing. You can find them on Ebay – pretty expensive. Do you read Erwin Puts?

        • I recall all of the second hand ones online disappeared not long after I posted the review! I’m sure Erwin is good technically, but I prefer to use the hardware and make up my own mind about whether or not it fulfills my needs.

          • Martin Fritter says:

            Puts is for retro-look types, I think. Or Leica “back catalog” folks. For example, if I want to look into R-mount lenses, I go to him. Anyway, I like his writing.

  4. Bill Walter says:

    There are very interesting articles Ming. I found that sometimes lens imperfections can be great. Like the interesting field curvature on my Zeiss ZE 28mm f2 when shot at f2. But this same lens when shot at f8 or f11 has excellent edge to edge sharpness. That’s a nice combination. The lenses that I find to be the most overrated are the fast lenses (like 50mm 1.2 or 85mm 1.2) that aren’t sharp wide open. Some photographers defend such lenses by saying that they have “character” when shot wide open. To me that’s just rationalizing poor performance. If I’m going to cough up a large sum of cash for such a fast lens, it darned well better perform well at it’s maximum aperture. If an expensive 50mm 1.2 performs poorly at 1.2, I’d prefer a 50mm 1.4 or 50mm f2 that has better IQ and costs a fraction of the price.

  5. About that “signature” and “indefinable quality”, for example, I’ve found the rendering of the Otus 55mm extremely attractive whilst finding highly attractive the value proposition of the Sigma ART at a price point of roughly 1/4. The smart, rational buyer psychology heads straight to the Sigma without hesitation.
    Then I look again at samples of how these lenses render and it niggles me that Otus images regularly appear more artful than the ART’s. And I believe there could be lots of skewing reasons for this experience including sample image selection, self-selection of buyers, and etc., that are all layered on top of the authentic differences in the products.

  6. gary bliss says:

    Ming —

    The authoritative and comprehensive nature of your comments makes this pair of posts one that i wish every photog would read and ponder, particularly the every verbose internet commentariat . . .

    May i add one tiny additional detail that i have found important in over, ah, well, too many decades of shooting as applying to zoom lenses? For lack of a better term, i would call it “consistency” of optical performance at a given aperture across the focal length range. I find it immensely frustrating to use zooms wherein i must constantly be aware of whether the lens is in the “good” range of focal lengths for the aperture desired. ALL zooms vary to some extent obviously, but for me, poorer ones degrade to the point that it is artistically relevant.

    This particular flaw seems endemic to almost all variable aperture (and, thus, cheaper and lighter) mid-zooms and tele-zooms; fortunately there are few exceptions that are, to me, worth seeking out.

    • This is a very good point, actually. Few zooms perform well at all apertures and focal lengths – mostly because it’s very difficult to make the necessary optical corrections for say both wide and telephoto, or moderate telephoto and extreme telephoto. Zooms for smaller sensors tend to be much better as fewer corrections have to be made (e.g. the Panasonic 12-32/3.5-5.6 and 35-100/4-5.6 collapsible compact zooms for M4/3 are really excellent, but nothing similar exists for FF). By the time you get to the other end of the market – medium format – there are very few bad lenses even if they have to be made variable aperture as a size compromise; the Hasselblad 35-90/4-5.6 is excellent, but given its current size – I’d hate to imagine it as a f2.8 constant aperture… 😛

  7. Hi Ming, you mostly touched on optical qualities. There are other items high on my list when evaluating what makes a good lens for me:

    1. Size and weight (High importance)
    2. Weather sealing (Med importance)
    3. Filter size (Low to med importance).

    Size and weight are high on the list of what’s important to me when selecting a lens since I like to shoot landscapes that require me to pack gear and hike. Fortunately, there are plenty of options available since fast glass with autofocus are not requirements for capturing high quality landscapes, but there have been cases where I’ve chosen ‘lesser’ lenses over an optically superior contenders based on weight and size. Weather sealing is also a factor that I consider since shooting outdoors a lot.

    Filter size is another consideration. Adapter rings mean forgoing lens hood; square filter holders require more care and add to setup time. Some UWA lens cannot easily accommodate a filter, even though they’re used for landscapes where more shooter is more likely to want a filter. As much as possible, I try to stick to lenses that accept a 77mm filter thread and avoid lenses that cannot accept filters, or that require very large & expensive filters.

    • Light, cheap, small and optically poor doesn’t necessarily mean ‘good’; if you’re a extreme hiker and it’s the difference between a photo or not, then it’s a better lens for you than a higher resolving but heavier one. But given that most people are aware of their own personal priorities and I’ve never had an email asking if one lens is lighter than another, yes, you’re right, I focused on optical qualities since there seems to be a lack of knowledge in that area…

      • We’re in agreement — by no means was I suggesting light, cheap, small, and optically poor are good! If I’m going to pack somewhere for sake of shooting a landscape, rest assured optical quality is important to me 🙂

        By way of example, I was on fence between Nik 20/1.8 and Zeiss 18/2.8. I’ll give Zeiss benefit of the doubt and assume it has marginally better optics (people will argue this), but for me it came down to weight and size. The 20/1.8 is almost 1/2 weight and takes 1/3 less volume. The 20/1.8 also adds autofocus and is about 1/3 the cost, although these were not primary considerations fo me (but certainly nice!).

        Anyway, thanks for another excellent post. Not trying to detract from the more relevant aspects of your post, which is to focus on optics and which was a good read for me.

        • The tradeoffs sometimes aren’t obvious, too. If you’re going to use the Nikon and Zeiss mostly at f2.8-4, where the Zeiss is just starting – the Nikon is well into its stride and between 1.3 and 2.3 stops down from maximum already (and thus the optical comparison changes quite a bit).

%d bloggers like this: