[Class A]

While comparisons of lenses in a technical sense are all well and good, they could never show the pixie dust because such dust, by definition, is not amenable to testing.

I don't think that that's true in the same way the phrase "what pixie dust is cannot be described with words" isn't true. Just because many have difficulties to capture what exactly that quality is, it doesn't mean that the quality cannot be captured in principle.

Something to ponder about: Did Pentax just luck out with the Ltds lenses? I think a very good lens designer -- given the permission to do so -- can design a "pixie dust" lens on demand.

[Clinton]

You will be told that only the FA Limiteds have "Pixie Dust".

And a few FA* and A* lenses

[GlennG]

I think a very good lens designer -- given the permission to do so -- can design a "pixie dust" lens on demand.

I'm with this thought as well, though I sure would be interested in knowing what a lens maker would actually DO, proactively, to create a pixie dust lens, and I sure would also like to see that translated into specifications. (Or, is this one of those corporate "proprietary" things, like the formula for Coke, or the "11 secret herbs and spices"?) I have a hard time believing that the generally accepted pixie dust lenses are due to luck.

[rparmar]

Just because many have difficulties to capture what exactly that quality is, it doesn't mean that the quality cannot be captured in principle.

Imagine a set T of all lens characteristics that could be captured exactly in a technical sense. Outside this, but including it entirely, is O, the set of all observable lens characteristics. O - T = P, the pixie dust.

If one believes everything visible can be captured technically during the design process, then O = T and P=0. But to believe this is also, in my interpretation, to believe photographers have (or will have, in some ultimate future) no place in photography. I am unwilling to make that statement. So I believe in pixie dust.

Something to ponder about: Did Pentax just luck out with the Ltds lenses?

No.

I think a very good lens designer -- given the permission to do so -- can design a "pixie dust" lens on demand.

Yes. Lens design is (mostly) a science that involves a compromise between optimising different optical qualities within the constraints of form factor, money, physical properties, physics, etc. Choosing which optical qualities to optimise and which to allow to deviate further from the optimum is part of the process.

[EricT]

How about here:
https://www.pentaxforums.com/forums/pentax-slr-lens-discussion/87796-official...lens-list.html
Wide Open: Pentax 31 Limited & Leica 35 Summilux ASPH | STEVE HUFF PHOTOS
FA 31mm F1.8 - ClubSNAP Photography Forums
Pentax FA Limited & EOS 5D!! - FM Forums

Pixie dust abounds...

This might shed some light on the 31 vs 35 L series comparison:
Re: FA 31 Ltd on Canon 5D Mk II - full format dSLR with FA 31 Ltd: Pentax SLR Talk Forum: Digital Photography Review

And the 77 and other 85s:
Re: Pentax FA 77mm f/1.8 Limited Review: Pentax SLR Talk Forum: Digital Photography Review

Thanks again, Ash! I'm a bit surprised with the Leica comparison, I can swear I've seen sharper shots with the 31mm wide open. Don't want this to turn into a FF vs. APS-C thread, but the limited castrations for the 5D are a bit sad, it's a real pity Pentax doesn't offer a full frame body for these lenses.

BTW, the Eric in that last dpreview thread is actually me, once again asking for side-by-side comparisons Don't get me wrong, I would really like a 77mm limited. It costs $1350 here, though, so it won't be until my next trip to the US!

[EricT]

IMO on full frame the FA31mm f/1.8 Limited easily beats the canon 35L in regards to flare tolerance and across the frame sharpness. The 35L has a higher peak sharpness but the corners are quite weak on full frame. In a recent non-scientific test I did with the full frame Leica M9 with a Leica M > to pentax K mount adapter from novoflex I tested the 35mm f/2 summicron ASPH, against FA 31 f/1.8 Limited and the Pentax lens comfortably held it's own though the leica lenses I tested it against produced results that consistently showed more contrast*. In a test against the Leica 35mm f/1.4 summilux ASPH the FA31 had to run for it's money but I can safely say the FA31 and the summilux performed identically at f/4 an over. The FA31 was visibly smoother than the Pre-ASPH Leica 35mm f/1.4 summilux in terms of bokeh production.


*I suppose this could be chalked up to the focus flange differential rather than any deficiencies in lens coatings.

Thanks for that comment about the 35mm L, very interesting! Another thing worth noting is that the Canon L lenses are huge compared to the limited lenses.

[EricT]

You bet Simon. On all the FA ltds - often. With the DA* 16-50 quite a lot also.

I get this with my 16-50mm as well, but you rarely hear "pixie dust" comments about this lens. Of course, maybe I'm just missing them in between all the SDM threads

[EricT]

I don't think that that's true in the same way the phrase "what pixie dust is cannot be described with words" isn't true. Just because many have difficulties to capture what exactly that quality is, it doesn't mean that the quality cannot be captured in principle.

Something to ponder about: Did Pentax just luck out with the Ltds lenses? I think a very good lens designer -- given the permission to do so -- can design a "pixie dust" lens on demand.

I remember reading articles written by the designer of some of the FA limited lenses. It was very interesting, and showed how the thought process of the designer was and what compromises he had to make to get the intended performance. Basically, if I remember correctly, he sacrificed edge-to-edge sharpness and CA control to get more pleasing out-of-focus areas. Since this is hard to measure in tests, I don't think a lot of modern lens designers dare to make such compromises anymore. I would be happy to be proven wrong, though!

Here are the links to the papers, can't find the translations right now, though.
Science Links Japan | Lens technical report. SMC Pentax FA77mm, F1.8 Limited and FA43mm, F1.9 Limited.
Science Links Japan | Trial use report. SMC Pentax FA77mm, F1.8 Limited and FA43mm, F1.9 Limited.

[rparmar]

I am working on an blog article based on those design papers. Been too busy with my teaching to get it done.

[germar]

I have no Ltd's, no *'s, no Pentax glass better than SMC-Tak, so I haven't experienced alleged Asahi pixie dust.

May I suggest two Taks that I believe may contain some early pixie dust:


Super Takumar 85mm 1.9

SMC Takumar 135mm 2.5 (part number 43812)

Both of these might not have the full blown 3d rendering of the FA limiteds, but great sharpness, buttery bokeh, and that famous Takumar color rendering/contrast make them early pixie prototypes in my humble opinion.

[les3547]

I see these terms used quite a bit when reviewers describe their lenses here - quite often describing the qualities of their prime high-end limiteds. Are these terms defined, quantified, or otherwise measured? If I were to examine a spec sheet for any lens, how would I tell whether it has rendering or pixie dust? Thanks. Glenn

I've been pondering that question for a long time. Before I invested in Pentax, I shot with Leica cameras and lenses (film) back in the '80s and '90s. Before I owned them I had Nikon gear. I switched to Leica because their lenses produced something I couldn't define. But knew I liked it, and so wanted to find it again using Pentax.

I haven't owned any FA LTDs yet, but I have hopes of at acquiring the 31mm soon to see if it has what I'm looking for. I did own the DA 35mm LTD and the DA15mm LTD, but sold them both. I know I asking for trouble by saying this, but I found them lacking the kind of pixie dust I'd had with Leica.

That's when I discovered the Voightlanders, and realized they render the unique sort of IQ that appeals to my tastes. But what is that quality?

Telling my wife about it, I said it reminds me of the difference between tube amplified music and solid state amplification. In the case of the DA LTDs I'd owned, they did everything right -- sharpness, color, etc.-- yet there was an "edge" to them that seem a bit stark compared to the Voightlander lenses, which yield all the detail of the picture, but in a more laid-back way (sort of like tube-amplified music).

Still, what is that quality? My current theory is based on how we see three dimensionality. I mean, reflected light from objects is reaching our brain and translated in a 3D way. That's how we experience reality (visually) so that's what we expect in a picture.

I think some lenses are able to capture light fall off, shadows, and highlights of a scene that are the same types of information which create 3D experience in the brain using our eyes. Using the stereo system metaphor, when a note is played live it has a natural duration from the initial strike of the note (called attack) and then a natural time it lingers, known as "decay." Good music systems are able to capture much of the note decay (if it's on the recording), while poorer systems may get little. If you are an experienced listener, music that is too much "attack" seems harsh, although those who like music that way may describe it as "detailed" or "crisp."

Similarly, it seems that lenses sometimes get defined primarily by sharpness, but for me that isn't enough. I want the subtle nuances of three dimensionality that naturally "soften" the edges of details, just like my eyes see things. Sharp and 3D . . . one can imagine it is difficult to achieve in one lens.

Pictures I've seen taken with the FA LTDs seem to have it (at least the 31mm and 77mm, I haven't seen enough 43mm pics yet), some Zeiss lenses appear to, I saw shots from a Leica lens (35mm I think) adapted to the Pentax that were amazing, and as I said, I believe several of the Voightlanders have it.

[Ash]

I too was sceptical about the 43 ltd amidst all the faster fifties around, but after going from a few different fast fifties to the 43 I've become a believer. The FA ltd lens club and that 43 is spoiling me thread offers a number of fine examples this little gem can do. It really is a fine piece of engineering and the f/1.9 maximum is not much of a limitation IMO.

[Class A]

O - T = P, the pixie dust.

I don't agree with this definition as it would associate pixie dust also with bad lens properties that cannot be captured exactly in a technical sense.

Also, "pixie dust" for me doesn't even include all technically non-tangible positive lens properties but only a specific subset of them.

Furthermore, it makes a big difference who it is who cannot capture lens properties "exactly in a technical sense". I'm not sure whether your set "P" isn't very, very small for an expert lens designer.

In particular regarding "pixie dust" (as I define), it cannot be completely in "P" for expert lens designers, because then they'd be completely in the dark as to where to find it. I believe that lens design involves trial and error in addition to planning but if you don't have at least an educated idea which dial to turn then you have next to no chance of creating a special lens on demand.

I remember reading articles written by the designer of some of the FA limited lenses.

Yes, I read those (written by Jun Hirakawa) as well. As you say, he mentioned (among other things) that he allowed the FA 77 to have a field curvature that isn't as planar as it could have been in order to optimise rendering.

I switched to Leica because their lenses produced something I couldn't define.

It is often referred to as "The Leica look". For people who do not know where it is coming from, it is pixie dust. I believe, however, that lens experts know pretty well where that look comes from. If I'm not mistaken, part of it has to do with what MTF value a lens is optimised for.

It is a physical fact that you cannot have optimal resolution and contrast at the same time (-> Heisenberg uncertainty principle). It is up to the lens designer where to tip the balance to. Whereas Zeiss typically goes for resolution, Leica typically goes for contrast.

Again, if the Leica engineers didn't know how the pixie dust of their lenses correlates to technical principles then they wouldn't be able to "dust" lenses in a consistent manner.

[Digitalis]

Zeiss typically goes for resolution, Leica typically goes for contrast

Zeiss also is less inclined to use Aspherics or ED glass in their lens designs, whereas Leica will use whatever it takes to produce lenses with higher contrast.

A similar comparison can be seen amongst audiophiles and the war between Tubes and Transistors.

[les3547]

It is a physical fact that you cannot have optimal resolution and contrast at the same time (-> Heisenberg uncertainty principle). It is up to the lens designer where to tip the balance to. Whereas Zeiss typically goes for resolution, Leica typically goes for contrast.

Again, if the Leica engineers didn't know how the pixie dust of their lenses correlates to technical principles then they wouldn't be able to "dust" lenses in a consistent manner.

I wouldn't venture a guess about how (technically) it is achieved. I do know post processing includes significant enhancements for resolution and contrast, and so far I've not been able to recreate "pixie dust" in a shot with a lens that isn't on the list.

To me that suggests the information isn't there.

Is my hypothesis plausible . .. that the design of some lenses allow them to capture certain 3D features that render a shot more life like (based on how our brain perceives) than other lenses? Isn't a common superlative applied to "pixie dust" lenses that they impart a sense of three dimensionality?