[rdenney]

Several threads have discussed the bokeh of lenses in this focal range for the 645. Since I have an interesting selection of lenses in this range, including some that are usually thought to be world-class in the bokeh department, I ran a (sort-of) controlled comparison. This is not that rigorous a test, but maybe it will give an impression of how these lenses render out-of-focus details.

First, some clarity. Bokeh is about the quality of that rendering, not the quantity of it. The longer the lens (if subject distance is preserved), and the wider the aperture, the blurrier those details will be. But that is not bokeh. Bokeh effects fall into several categories:

• Out of focus highlights are rendered with bright edges. This often results from optimizing for apparent sharpness at wide apertures, which has been a hallmark of Japanese lens designs in past years. (And that generalization demonstrates that all generalizations are ultimately false. But the exceptions to it are rare enough to prove the rule).
• Those highlights are rendered with a neutral disk that has a sharp edge but that is not brighter than the rest of the disk.
• Or, the highlights are rendered with a faded edge. This is usually the result of slight (perhaps intentional) undercorrected spherical aberration at wide apertures. (Soft-focus lenses achieve the effect using spherical aberration.)

Other attributes can also be apparent. Sometimes, out-of-focus backgrounds have a swirly effect, where they seem to arrange in a circular pattern around the image center. Bright-edge bokeh may create false detail in the background as fuzzy spots overlap and create aliasing patterns, making it look closer to focus than the details would be without those overlaps. This undermines the smooth transition from sharp to fuzzy with greater distance from the focal plane. That smooth transition is what creates the illusion of three dimensions, and false detail undermines that illusion.

It should be noted that my own preferences lean toward smooth, faded-edge bokeh with smooth transitions. But that is a subjective determination. Others prefer different rendering to achieve their artistic intentions. So, there is no better or worse in this comparison. I just observe the effects that I see, and it's up to you to determine what you like, despite the opinions I may express.

The test camera is a 645z, mounted on a Gitzo tripod. The sunny conditions made the tripod unnecessary--even at ISO 100, shutter speeds were quite high at the apertures used in the test. But I wanted to keep the camera aimed to the same focus point between shots, so I used the tripod for convenience.

Here's the lineup:

• Pentax 645 FA 150/2.8 with Pentax shade.
• Pentax 67 200/4 (Pentax adapter) with integral shade (the front element is very deeply recessed). As far as I know, this is the latest design.
• Pentax 645 FA 200/4 with Pentax shade.
• Pentax 645 A 80-160, at 160, with aftermarket rubber shade.
• Pentax 67 165/2.8 (Pentax adapter) with slide-out integral shade.
• Carl Zeiss Jena MC Sonnar 180/2.8 with Hasselblad compendium shade, in Pentacon Six mount with a Hartblei adapter. This is a classic Sonnar design from the 50's that has only been tweaked slightly from the original Olympia Sonnar. This Sonnar design uses the original very thick elements not found in later western Zeiss Sonnars. This lens was made in 1978, probably near the peak of Zeiss Jena quality control, when they were almost up to western standards. This is the later model with multicoating and the stepped barrel, but not from the last few runs that had four and five-digit serial numbers. This is not the sharpest lens I own by any means, but the rendering is what makes it special. It's a fast lens that is most interesting at wide apertures.

(All the arrangement pictures were made using a Canon 5DII.)


Test Setup 1

In this arrangement, the focus target was an iron boot last (which I now realize I forgot to bring back into the house). That was about the only thing I could find that wouldn't move in the breeze, but was tall enough to poke up into view in the bed of daffodils.



All these images were made at f/4, except for the zoom, which only goes to f/4.5.


FA 150


P67 200


FA 200


A 80-160


P67 165


Sonnar 180

I was looking at these, scrolling through them in DxO Photolab, and my wife was looking over my shoulder. She did not know which lens was which, and doesn't know enough about what I was testing to be able to tell based on magnification. But she said, "That's the Sonnar" when it popped up. The rendering is that characteristic, even to a non-techie. The P67 200 was close, and much smoother than the 645 200. The fast 150 was not as nice. The P67 165 had the swirly thing going--again, not my favorite.

Before leaving this setup, I compared the three f/2.8 lenses shot wide open. Here are those images:


FA 150 at f/2.8


P67 165 at f/2.8


Sonnar 180 at f/2.8

The Sonnar is the clear winner here, per my own preferences. Also, it completely lacks longitudinal chromatic aberration which is evident in the two Pentax lenses when viewed at 100%.

Test Setup 2

In this test, I was wondering what a busier background would look like with a longer subject distance. The focus target was my famous River Birch, and my Azaleas in the background are just in the bud stage. The longer distance brought the background closer to the the focus plane with respect to the camera, and this is a bit of a bokeh torture test.


FA 150. Lots of false detail and a busy rendering.


P67 200. Still a bit busy, but much nicer to me.


FA 200. Bright-edge bokeh is visible here, and it's not as nice as the P67 200.


A 80-160. Not all zooms have crispy bokeh, but this one does.


P67 165. Like the P67 200.


Sonnar. Best of this batch, though still not completely smooth with this subject material.

Test Setup 3

These shots are all at f/11, so I'm not looking at out-of-focus rendering, but rather landscape rendering. So, does good bokeh mean bad landscape images, or the reverse? Is there correlation? Let's see.

Here's the setup. The camera is looking past the house to the Dogwood and Japanese Maple trees in the side garden.



Here are the images. I'm not zooming into these--they were all shot at f/11 and were sharp enough even at 100% to be wholly usable even for large prints. I'm more interested in the overall rendering, color, and contrast.


FA 150


P67 200


FA 200


A 80-160 at 160


P67 165


Sonnar

Not a lot of difference here, but the Sonnar is my least favorite of these. Go figure.

Test Setup 4

Back to the front garden and those Daffodils. This time, I was aimed a bit differently to put my Ford Expedition in the background. My intention here is to provide some specular highlights and technically crisp edges to evaluate. The focus point is still that iron boot last.



Images:


FA 150 at f/4


P67 200 at f/4


FA 200 at f/4


A 80-160 at f/4.5


P67 165 at f/4


Sonnar at f/4

No clear winner here, though there are a couple of losers. I think the P67 200 might be the match of the Sonnar, or even edge it out. For bokeh, the P67 200 is clearly a better performer than the FA200 for the 645, though the 645 lens is sharper at optimal aperture and MUCH easier to carry around.

Finally, a series of pictures of the lenses mounted on the camera to get a sense of their bulk. Here's where the Sonnar comes in as the BEAST.


FA 150


P67 200. This is much bulkier than the 645 200, but it's not that heavy.


FA 200. Small, light, sharp at optimal aperture, but NOT a bokeh monster. Most of what you see is lens shade.


A 80-160. A bit beastly, being denser than either 200.


P67 165. Larger and heavier in this older version.


CZJ Sonnar 180. Wider rather than long, and very heavy. The compendium shade is NOT the reason for the weight. This Hasselblad shade is light and excellent. The lens comes with a hard plastic screw-in shade (for the 86mm filter ring), but the compendium shade is much more effective for a lens often used with studio lights.

So, as I have concluded in the past, bokeh is subjective but it's also real. The Sonnar makes a superb portrait lens, and comes closest to the 3D effect we often seek with the use of selective focus. But the P67 200 is no slouch.

Rick "respectfully submitted" Denney

---

Last edited by rdenney; 04-19-2020 at 04:55 PM. Reason: Typos

[RICHARD L.]

Very nice work, sir ! Shorter lenses tend to create fuzzy backgrounds with lots of detail, not really a uniform canvas like a studio background would. Only longer lenses or beasts like Nikon's 200 mm f/2 tend to produce this kind of detailless backgrounds.


P.S. The lens that impresses me the most is the 67 Pentax 200 mm f/4 in all of your trials. I have one that hasn't been used in years, maybe I should give it a try soon.

Regards

[Sandy Hancock]

Nice work Rick.

I'll just have to soldier on with my second rate FA150/2.8 and FA80-160/4.5

[HarisF1]

I think a direct test should also include wide open results as that's where the appeal of the f/2.8s would show.

[rdenney]

I think a direct test should also include wide open results as that's where the appeal of the f/2.8s would show.

Test Setup 1 included that.

Not that I think they change my conclusions.

Rick “who has them for Setup 4 also, but unposted” Denney

[kaseki]

I'd give +24 likes if I could.

[Ed Hurst]

What meticulous, interesting and generous spirited work! Thank you sir.

Ed "Ed Hurst" Hurst

[rdenney]

Thanks guys.

Rick “enjoying (only) some aspects of the lockdown” Denney

[HarisF1]

I'm very surprised at how poor the FA 150 seems to be at actually creating dreamy bokeh. It's very clinical in nature. I've got one but I'm just a collector, not much of a user so I haven't really worked with it too much.

What I find amazing is the 105/2.4

[barondla]

Very informative test. All the lenses have their own particular character. For the Pentax primes, this is a little surprising. Always figured they would have certain optcal formulas that would be scaled up or down to make a 150 or 200 lens. Not totally different for 150, 165, and 200 mm.

Thanks for all the work put into these tests,
barondla

[rdenney]

Very informative test. All the lenses have their own particular character. For the Pentax primes, this is a little surprising. Always figured they would have certain optcal formulas that would be scaled up or down to make a 150 or 200 lens. Not totally different for 150, 165, and 200 mm.

Thank you. But I'm not surprised. Even the Pentax lenses have quite different provenance. Summarized from the lens review pages:

• 645 FA 150/2.8: A modern design in seven elements and seven groups with internal focusing, first appeared with the FA series and the introduction of the 645N camera, so it's no older than 1997. It would have benefitted from fully optimized computer-aided design.
• 67 200: the later SMC-Pentax 67 version with a new optical design, dating from the late 80's. But it's still a traditional design in five elements and four groups.
• 645 FA 200: Like the 150, a modern design in six elements and five groups with internal focusing, developed for autofocus, and still (officially, at least) in the line.
• 645 A 80-160: I bought he manual-focus version because it's the same optics as the FA version, but a lot cheaper. I don't use zooms in this range that much, and the right deal came along. Because of the optics, it's an older design, probably from the early 80's, which would benefit from some computer optimization. Like all zooms, the design is complex.
• 67 165/2.8: This is an older SMC-Pentax 6x7, which puts it in the middle production period, probably from the late 70's or early 80's. It is a highly corrected six-element, five-group design, using early computer optimization.

So, we have Pentax lenses from four distinct periods, probably designed for different price points and making use of different design tools and resources.

Rick "noting how multicoating made more elements possible, and then computer optimization made still more possible" Denney

[barondla]

Thank you. But I'm not surprised. Even the Pentax lenses have quite different provenance. Summarized from the lens review pages:

• 645 FA 150/2.8: A modern design in seven elements and seven groups with internal focusing, first appeared with the FA series and the introduction of the 645N camera, so it's no older than 1997. It would have benefitted from fully optimized computer-aided design.
• 67 200: the later SMC-Pentax 67 version with a new optical design, dating from the late 80's. But it's still a traditional design in five elements and four groups.
• 645 FA 200: Like the 150, a modern design in six elements and five groups with internal focusing, developed for autofocus, and still (officially, at least) in the line.
• 645 A 80-160: I bought he manual-focus version because it's the same optics as the FA version, but a lot cheaper. I don't use zooms in this range that much, and the right deal came along. Because of the optics, it's an older design, probably from the early 80's, which would benefit from some computer optimization. Like all zooms, the design is complex.
• 67 165/2.8: This is an older SMC-Pentax 6x7, which puts it in the middle production period, probably from the late 70's or early 80's. It is a highly corrected six-element, five-group design, using early computer optimization.

So, we have Pentax lenses from four distinct periods, probably designed for different price points and making use of different design tools and resources.

Rick "noting how multicoating made more elements possible, and then computer optimization made still more possible" Denney

Good points. Hadn't considered the different ages of the lenses. Obviously, as you point out, AF and internal focus caused optical changes. Affordable aspherical elements could also be another game changer.

But, looking at FF & APS-c lenses, it's interesting that they've always (last 20 years or so) made great 300mm and average 200mm lenses. Their 400mm lenses don't have as many enthusiastic followers either. It would seem same class teles, designed at the same time, would use the same optical formula. Not convinced they do.

Another example of this in 645 land is the fact that all the 35mm lenses seem to be from good to great. While the 45mm (which should be easier to design ) are meh. Just strange.

Thanks,
barondla

[rdenney]

Good points. Hadn't considered the different ages of the lenses. Obviously, as you point out, AF and internal focus caused optical changes. Affordable aspherical elements could also be another game changer.

But, looking at FF & APS-c lenses, it's interesting that they've always (last 20 years or so) made great 300mm and average 200mm lenses. Their 400mm lenses don't have as many enthusiastic followers either. It would seem same class teles, designed at the same time, would use the same optical formula. Not convinced they do.

Another example of this in 645 land is the fact that all the 35mm lenses seem to be from good to great. While the 45mm (which should be easier to design ) are meh. Just strange.

Thanks,
barondla

Scaling designs may be the problem. Doubling the size of the elements of a lens in all linear dimensions results in a focal length twice as long but with the same focal ratio (f number). But the thing to remember is that it also doubles the effect of any design-related aberrations.

Thus, perhaps the conventional 200's were thought to be sufficient, but the four-element short telephoto magnified aberrations too much at 300mm and up. So, they developed a new design from the ground up, following a much more optimized process and making use of special glass, resulting in the 300 ED lenses. Or, they ratcheted it up, with a fresh 300 design that they backed up to 200, and then realized they could improve the 300 again as customers and competition became more demanding. The 300's have more elements and technology by far than the 200's. I think this is plausible given the very good performance of the 200's that were developed at the time (and that performance is very good--I'm a big fan of both the P67 200--the later design--and the 645 FA 200 when making sharp images at optimal apertures, which is most of what I do.) But to get that level of performance at 300, they needed more advanced (and expensive) technology. Then, they scaled up (approximately) the 300's to make the 400's. My 645 400 ED is quite good, but maybe not quite as good as the A* 300 ED.

For wides, the process is reversed, being a reversed-telephoto design, and shorter lenses require more magnification in the corners to provide rectilinear projection. They use one or more negative elements in front of a conventional gaussian design to accomplish that. That greater magnification also magnifies aberrations, so the shorter the lens, the bigger the problem. The A35 is a different design than the FA, and by all accounts really got it right. But I think it's still a later lens than the 45, or perhaps they thought the 45 was sufficient but needed a wholly different design for the 35. I don't think it's just the 45mm focal length--the P67 45mm lens is stunningly good; at least mine is. I expect the P67 45 might be a scaled up 35, but I have not at all studied the diagrams to investigate that suspicion. But we do have to remember that until the 70's, 50mm was very wide for a medium-format SLR. Computer optimization really opened up possibilities for retrofocus wide-angle designs.

All of this is pure speculation, of course. But even though lens performance has come a long way with better coatings, aspherical elements, low-dispersion glass, and computer-aided design, not all lenses have been updated to the latest of everything unless there was a specific reason to do so. Maybe Pentax didn't see the need to update the 45 because the 45 end of the 45-85 was so good (and it is really excellent).

So, why is the Sonnar so good? We can never discount serendipity. When Bertele designed the first Sonnar, he faced strong limitations. Optical designs were carried out at large scale using ray-tracing techniques, in the hopes that the lens would resolve detail 1/1500th the height of the frame. In 1929 when Bertele was working on it, he did not have coatings at his disposal. So, he was constrained to using few elements or cemented groups. And he was limited to spherical shapes, and had a relatively limited range of glasses available in the Schott catalog compared to later years. Despite those constraints, he came up with a brilliant design. The first Sonnar, a 50mm f/1.5 for the Contax, used seven elements in three groups. It was really an enhancement of a triplet design. Was Bertele lucky, or did he enjoy a flash of insight? I know I've had a few ideas in my life that worked out stunningly well, and lots of other that...didn't. The ones that didn't were adequate to the task, but lacked the brilliance of the few. The only problem with the Sonnar was that it required close proximity to the film relative to focal length. So, it worked fine as a normal on a rangefinder, but interfered with a reflex mirror at shorter focal lengths. So, it became a short telephoto, though it is not, in fact, a telephoto design (the rear node is not in front of the physical center of the lens). The design was a bit simplified to five elements in three groups, and the first example was the 180mm f/2.8 Olympia Sonnar famously used by Leni Riefenstahl to photograph the 1936 Olympic Games in Berlin. But even though it was fast, there remained a hint of spherical aberration at wide apertures, which contributed to the stunning bokeh and gave the design its legendary status.

Interestingly, Nikon made a 180/2.8 lens that copied the traditional Sonnar design, back in the days of the F and F2. I own it, and use it from time to time on my Canon with an adapter. That lens is also legendary. The Nikkor 105 is the same way. I sought out both, despite my high opinion of the Canon 70-200 f/4L zoom, even in the bokeh department.

But Zeiss Oberkochen long ago modified the Sonnar design to be more like an Ernostar, with much more space and much less glass. That takes advantage of better coatings and it's much lighter, but it lacks the legendary rendering in my view. (It's also not that much sharper according to tests I've conducted, but with too few samples to really know.) I think the motor that would be needed to autofocus the traditional Sonnar would require a car battery.

Rick "old memories" Denney

[w2ck]

Hi,

I really like the Sonnar used in this test. But, then, I have one of those Nikkor 180/2.8 as well as a 105/2.5 left over from my F2. I converted both to AI for my FE and FA bodies long ago. And, I still use them rather than their newer counterparts, these days on a Df. So, maybe there is no secret as to why I like this one shown here.

Stan