PHOTOOG Photography writings by Olivier Giroux

25Apr/080

Lens Connoisseurship

I don’t quite fancy myself as a lens connoisseur, but I understand that there are people with the fine tastes that can tell a great lens from the rest. Like wine. And just like wine, there’s ten thousands pretenders for each master. I can’t tell you how many lens reviews I’ve read that claimed “Good Bokeh” when the supporting evidence makes my eyes hurt by just looking at it.

I’ve so far sought to acquire two skills which I believe are part of a lens connoisseur’s repertoire.

Recognizing Good Bokeh

Quick facts:

  • Bokeh is Japanese for “broken”, means the texture of the out-of-focus regions of a photograph, more info here (http://en.wikipedia.org/wiki/Bokeh)
  • Bokeh can look smooth or busy/nervous/jittery, depending on the correction of spherical aberration in the lens (http://en.wikipedia.org/wiki/Spherical_aberration).
  • Bokeh is also affected by Astigmatism (http://en.wikipedia.org/wiki/Astigmatism) but the relationship is less well documented (or I just couldn’t find the documentation).
  • Good Bokeh is either the Smooth kind (due to residual aberration in the direction of defocus) or the Neutral kind (perfectly balanced aberration control).
  • No lens can have Good Bokeh on both sides of the plane of focus unless it is Neutral Bokeh, but if you have to choose you’ll probably want Good Bokeh behind the plane of focus.
  • You can’t judge Bokeh from an image where (1) the defocused area has melted down completely and/or (2) there are no specular highlights in the defocused area.

As you can see there’s actually a rather mechanistic description of what Good Bokeh is, despite the mass inability to make the call on this. I think most people are confused between “Good Bokeh” and “Melted Background”. Even a lens with Bad Bokeh will give a creamy background if you’re hugging the subject in front of a featureless background at f/1.2. And of course it sounds cool to say your new favorite lens has “Good Bokeh”, especially if you bought it to melt backgrounds.

I’ve made a few other observations of my own that I’m not quite ready to mark down as “Bokeh Laws”…

  • Simple lenses with a few spherical elements do very well as a group. The megapixel race emphasizes complex lenses that resolve finer detail but have almost uniformly Bad Bokeh, the older Double-Gauss designs do much better.
  • More specifically, lenses with Aspherical elements tend to give Bad Bokeh. Even the ASPH Leicas. The best known Bokeh is probably found in the pre-ASPH Leicas.
  • Zooms as a group do poorly with Bokeh. If I had to venture a guess, I would say their flavor of spherical aberration control puts the Good Bokeh in front where it’s least likely to be useful. (…or it could be a conspiracy to sell fancy glass to connoisseurs!)
  • Macro lenses tend to do well, but at macro distances the “melting” effect is so important that it can easily hide Bad Bokeh.

But do you need Good Bokeh? That’s an entirely different question – like whether you need the best wine. Most lenses do not have it, yet most everyone (professionals included) live without it, usually thanks to “melting” at wide apertures.

I haven’t seen a scientific study of the topic, but I suspect that the Zeiss ZFs do very well on Bokeh. Most employ a moderate number of large spherical elements to achieve their goals – which to me seems like it’s conducive to Good Bokeh.

Reading MTF charts

MTF charts are an objective report of lens performance, a cold description of the ability of a lens to converge monochromatic light rays properly on top of one another. They most notably don’t say anything about chromatic aberrations.

They also don’t say anything about making beautiful pictures. Some people get mad when their favorite lens is criticized by MTF charts because they believe, rightly, that underperforming lenses can skill make striking photographs. They totally can, but that says more about the photographer than the lens.

In an MTF chart you will find information about:

  • Color contrast
  • Edge contrast
  • Astigmatism

You can then infer, that is guess, things like field curvature and whether spherical aberration is well controlled. You can also get a feel for the likelihood the lens has Neutral Bokeh from the astigmatism information.

MTF charts basically follow the simple “more is more” philosophy that any modern consumer can identify with, that is higher values are better. Each chart is made assuming the lens is used wide-open, that is at its worst aperture typically. Some manufacturers (e.g. Zeiss) also include MTF charts for the lenses closed-down a stop or two from full aperture.

On an MTF chart you will find at least 2 pairs of lines: one for color contrast (always higher) and one for edge contrast (always lower). Each pair describes contrast radially and tangentially from concentric rings of samples taken on the image plane. The vertical axis is the % of contrast retained from the source. The horizontal axis is the distance in mm of the sample ring from the center of the lens.

Note that the DX frames ends at 12mm horizontally, 9mm vertically and that its corners are at 15mm. The FX frame ends at 18mm horizontall, 12mm vertically and 22.5mm diagonally. Each MTF chart then goes from 0 to ~22.5mm so you can see the performance across an entire FX frame. Typically DX cameras benefit from a “sweet spot” effect with FX lenses by using only the best part of each chart.

Finally, if the radial and tangential lines of the edge contrast pair are separated by a wide margin, then the lens is astigmatic at that distance from the center, and the bokeh will not be Neutral (but could be either Good or Bad).

Here are a few examples:


Sigma 24mm f/1.8 EX


Nikkor 24mm f/2.8D


Zeiss 2.8/25 ZF

Note that Zeiss records both coarser and finer edge detail on their MTF charts. To compare with Sigma and Nikon, imagine a curve pair that traces halfway in-between the 2 lower curve pairs.


Nikkor 24mm f/3.5D PC-E

Let’s try to draw a few conclusions from these charts. I’ve purposefully chosen a difficult focal length to make this interesting.

    • The base reference for the comparison is the Sigma. The chart reveals its performance degrades almost linearly straight from the center, which sounds like the price it pays for its sup
      er-bright f/1.8 aperture. My feeling is that it has pretty significant light fall-off towards the borders/corners wide-open and this is what’s dragging down contrast. It’s also possible that it’s field curvature is causing this, which would be a more serious problem. The lens is clearly not astigmatic however.
    • The first Nikkor doesn’t completely outperform the Sigma, yet it’s 1.3 stops slower wide-open. On a reduced-frame DX camera, it would be hard to make the call between the two – I’d go either by speed or size/weight. On an FX camera I might prefer the Sigma, pending a look at chromatic aberrations.
    • The Zeiss presents itself as a different beast. Its center performance isn’t quite as good as either the Sigma or the Nikkor, but as soon as you move away from the center it is superior to both and draws much more consistently across the frame. Of particular note is the much higher color contrast curve, which is a trademark of Zeiss. Having read other reviews of this lens which rated it lower than most Zeiss ZFs, I suspect we’re seeing some astigmatism here.
    • Finally the second Nikkor is clearly something special. It’s another two-thirds of a stop slower, a full 2 stops under the Sigma now, but delivers class-leading color and edge contrast from its full aperture. The color contrast is on a level you would expect from a short telephoto – this is a most astounding achievement, which probably comes from the employment of Nikon’s newfangled “Nano Coating” in the PC-E’s manufacture. Edge contrast is also as good as it gets for this focal length.
    Of particular note is the long tail of the PC-E's edge contrast curve which indicates that the lens retains brightness and detail up-to and beyond the border. This is expected from a good tilt/shift lens which, when shifted, extends the image field far past the end of the chart to 31mm. The MTF chart doesn't tell us then how well it performs when shifted, we can only assume that the tail extends some distance and then falls down as with the other lenses, around the 31mm mark.
    If we had instead looked at long telephotos the MTF charts would show mostly flat lines, hugging the top of the chart. The discussion would have been a bit boring, but it tells you there is much less distance between the products there. Telephotos aren’t optically difficult to build – that’s my personal conclusion.


    Comments (0) Trackbacks (0)

    No comments yet.


    Leave a comment

    No trackbacks yet.