Lessons from a Legendary Lens Designer


About ten years ago, the photographer and lens guru Mike Johnston, the man who introduced the term bokeh into the English language, asked the following intriguing question: who makes the very best autofocuslenses that money can buy? He answered:

It's got to be Nikon or Canon, right? Each of these Goliaths, with their vast lens lines and cost-no object fast lenses and zooms, have won the battle of public opinion going away. So here's a shocker. The real answer may be Zeiss and Pentax! Zeiss, with the jewel-like little G lenses for the Contax G1 and G2, and Pentax with its little-heralded but lovely Limiteds.

And Johnston continues:

All three [FA Limiteds] are utter standouts optically. With the vagaries of personal taste taken into account, no lens, however deluxe, can be called the "best" for everyone, but the Limiteds are certainly among the best. Popular Photography in its March 2002 issue called the Pentax SMC-FA 31mm Limited one of the greatest prime lenses it had ever tested (the other two were the Voigtländer Heliar 50mm f/3.5 and the Nikon Nikkor 45mm f/2.8P Tessar-type. This wasn't clear in the issue itself, but I contacted the Editor, Jason Schneider, who confirmed it). Yet all things considered, the 77mm may be the best lens of the three. A nearly ideal short tele, the 77mm Limited is superb — contrasty, excellent for portraits wide open, with a truly beautiful, delicate bokeh that compliments the almost 3-D vividness of the in-focus image. Tops in its class? There are certainly a lot of great short teles out there. But I can't name an AF SLR short tele I'd put above it.

That’s high praise indeed. But another little known fact concerning these Pentax limited lenses is the philosophy that went into their making. That is a story in itself, one that is well worth the telling.

Two of the FA Limiteds, the FA 43 f1.9 and the FA 77 f1.8, were designed by legendary lens engineer Jun HIrakawa. According to Hirakawa, these two lenses were not designed “without being tied to preconceived notions.” In a short paper, Hirakawa explained the actual design philosophy behind the creation of these lenses in a short paper. Unfortunately, the available English translation is very poor: so poor, in fact, that at times it seems to contradict itself. For example, consider the following two sentences.

In regards to the specs, they vary little from what has been conventional wisdom up until now. Especially related to the focal length number, which has not existed until now but seems to have drawn interest from many...

If the specs of the lens vary little from conventional wisdom, why are their focal lengths unusual? Here we have an example of an obvious translation error. I suspect it should have read something along the following lines:

In regards to the specs, they vary in some respects from what has been conventional wisdom up until now. One particularly apposite example is their focal lengths, which has not existed until now but seems to have drawn interest from many...

Because of the difficulty of the translation and the liberal use of technical terms, I’ve decided to write a synopsis of the main part of the article, that which deals with some of the main conceptual ideas behind the two FA limiteds that Hirakawa designed. Where I have found contradictions or obscure passages, I have chosen the viewpoint most consistent with the general tenor of the article. In any case, I have provided, in PDF form, a copy of the original English translation, so anyone can compare them if they wish and decide whether I have taken unwarranted liberties. I have not reworked the entire article, only that portion starting with “The Trend of Aberration Correction”:

By not giving a priority to resolution, MTF measurements, and other numerical evaluations, the limited lenses achieve a degree of aberration correction unattainable by earlier lens designing practices. This enables us to manufacture a lens capable of producing images that leave a vivid impression on the mind. This is because, under current lens design practices, numerical evaluations of the subject plane take precedence over how well that lens depicts (or renders) solid objects. While numerical evaluations are important, they should not be given first priority when designing a lens.

As an example of this new attitude toward lens design, I will explain how we dealt with field curvature and astigmatism correction. An ideal of current lens design is to make the subject plane as flat as possible. Through specific techniques [detailed in the original article], field curvature is reduced, leading to a lens that will produce excellent scores on numerical tests. However, flattening the subject plane comes at a cost: pictures taken with such a lens will lack spice. With the Limited lenses, small amounts of field curvature were left uncorrected so that we could remove “astigmatic difference.” While this reduces the overall numerical score of the lens, without the “astigmatic difference,” the point of focus can be depicted with a more tactile, three-dimensional rendering, leading to a better overall image.

About Correction of Chromatic Aberration

The Limited lenses weren’t merely corrected for various chromatic aberrations. Focus points for each color and out of focus boundaries were also aligned, allowing for a gentle transition from the object in focus to the out of focus portions of the image. The corrections applied to the Limited lenses are a bit different from those normally used in the design of lenses.

Dealing with Inner-Surface Reflections

In regards to coatings, the well known super-multi-coating and the newer “ghostless” coating were employed in the Limited lenses to suppress ghosting and flare. In particular, the ghostless coating features excellent transmission in the red field, allowing the Limited lenses to be very effective in making vivid depictions of red subjects. In terms of the overall color scheme of these lenses, they still conform to common C.C.I. standards, so that the overall color scheme won’t change if the user switches to a non-Pentax lens.

Real PIcture Taking

As explained above, the design goals of the Limited lenses are difficult to evaluate on a numerical basis. No tests can measure the suitability of aberration correction on solid (i.e., “tactile,” three dimensional) photographic subjects or the effect of flare and ghosting on rendering. These things can only be evaluated by taking real pictures. With the help of professional photographers, we made qualitative evaluations of our designs.

The Limited Lenses are being developed with the ideal of high quality rendering in mind. They represent a break from preconceived notions concerning the importance of numerical evaluations in lens design. Instead, we wanted to make a lens that would produce beautifully rendered, unforgettable images.

Regardless of how “accurate” my rewording of the bad English translation may (or may not) be, I think the general message of the article is obvious: the Limited lenses were not designed primarily in reference to numerical tests. Much greater emphasis was given to the qualitative, non-measurable characteristics of lens performance. What lesson should we draw from this? Namely this: that it is how a lens performs in the field, the images it produces in actual use, that is critical in evaluating a lens, not how well it performs on numerical tests. There is way too much emphasis placed on how lenses and cameras score on resolution tests, as if such scores exclusively determine image quality! We need to emphasis the images produced by lenses and cameras, not just their numerical scores. Look, don’t calculate!