Finally, Some More m4/3 MTF Testing: Are the 40s Fabulous?

March 7, 2018 · Roger Cicala · in Geek Articles , Resolution Tests , Technical Discussions

It took me a bit longer to get back to m4/3 testing than I expected, but we have an interesting set of primes to look at today, the ‘about 45mm’ group, which includes several 42.5mm and a couple of 45mm lenses. Please don’t get all worked up about that small difference in focal lengths; it’s really not significant. One of the 45mm lenses is a 43.5mm, one of the 42.5mm is a 43mm lens. Plus if you (or your camera) does any distortion correction that may change a bit. Anyway, I really can’t imagine anyone going ‘wait, I need to change my 42.5mm to a 45mm for this shot’, so it seems a reasonable grouping.

For those of you who haven’t read our previous m4/3 tests, I want to emphasize a couple of things. First, we’re testing the lenses with no camera involved; all we’re evaluating is the optics. My interest is in the lenses, and the lenses only. Making lenses for a smaller imaging circle has some distinct theoretical advantages. Whether the lensmakers use those advantages for good and make better lenses, or for evil to cut corners and raise profit margins, well, that interests me quite a bit.

When you take a picture, you’re using the system (the camera and lens each add their limitations). Despite what many self-described experts say, it is very, very rare that just the lens or just the camera limits the output of the system. In practical terms, unless you have a really horrid lens or a 4-megapixel camera, both camera and lens contribute to the final output. (If you have both a horrid lens and a 4-megapixel camera then you don’t need to be reading this.)

One other point: with m4/3 lenses you can’t make broad generalizations about the name on the outside, because that doesn’t necessarily reflect who made the components on the inside. So if a given Olympus or Panasonic lens is superb (many are), don’t take that to mean all that brands lenses are pretty awesome (many aren’t). The name on the outside means “We paid the people who designed the lens, made the different components, and assembled it. Some of them actually get paychecks with our name on them, some don’t, and the ones who did this lens didn’t necessarily do that other lens.”

Let’s Meet Today’s Players

We have several lenses that meet the criteria of being around 45mm primes and being testable on our machine.

Because some of theses aren’t common renters, I wasn’t able to get a complete set of 10 tested for every model, as noted above.

I think the big question I have today is do they make the name longer because of some inferiority complex? If my theory is correct (it seems to hold for SLR lenses), then a graph of image quality should look like this:

If that holds true, then pretty soon we won’t have to do these tests. We’ll just go ‘that things got five words and six initials in the name, it’s gotta be horrid’.

Olympus M. Zuiko Digital 45mm f/1.8

At $400 this isn’t a cheap lens by any means, but it’s certainly a lot cheaper than some of those we’re looking at. At f/1.8 it’s not the fastest, either, and its seven aperture blades aren’t the most. But it is nice and small; 2.2 inches long and 116 grams in weight, and it does come in your choice of silver or black, so it’s got that going for it.

Olympus M. Zuiko Digital 45mm f/1.2 ED PRO

It’s priced equivalently to three Olympus 45mm f/1.8 lenses ($1200) but you get over three times the weight (410 grams) for your money, so that seems fair. Seriously, this lens is reported to be excellent, with a lot of design care for smooth bokeh, nine aperture blades, and reportedly excellent image quality.

Panasonic Leica DG Macro Elmarit 45mm f/2.8

This lens is probably not fairly compared to the others since it’s a 1:1 macro lens and not a wide-aperture prime, but well, it had to get tested somewhere. It’s a bit pricey at $800, and moderate in size compared to the others. It’s also an older design, and I didn’t expect it to be awesome.

Panasonic Leica DG Noctitron 42.5mm f/1.2 ASPH Power OIS

Of course, this is mainly about the Battle of the f/1.2 lenses, and Panasonic’s f/1.2 entry is the most expensive ($1,400), the heaviest by a hair (425 grams), and a clumsy hood design. But it has the Leica name on it, which means, well, nothing. But it does have the nice 9-blade aperture and the Power OIS system. So while we’ll compare it optically to the Olympus, your choice is probably going to come down to which body you shoot. And because I like to put my preconceived notions out there since they do affect my judgment, I want to not like this lens because it has both the longest name and the pretentious, yet meaningless, Leica badge on the outside; a Marketing Perfecta.

Panasonic Lumix 42.5mm f/1.7 ASPH Power OIS

Of course, if what you’re most interested in is having Power OIS, this gives you that, along with the lowest price of the group ($350) in a nice small package. I always root for the underdog, because, well, I’m cheap. So I’m going into this test hoping this lens will kick serious butt; or at least run with the big dogs.

Voigtlander 42.5mm f/0.95 Aspherical

OK, I stated on the last set of tests that I despise Voigtlander lenses. That’s because they simply can’t be repaired reasonably and parts are not available, so they are basically disposable lenses. But, I have to say, I was markedly impressed by the 25mm Voigtlander’s performance. And this lens provides by far the widest aperture at a middle-of-the-road $800 price tag. You pay a weight penalty because of the metal housing (and if you think a metal housing makes lenses more reliable, come spend a day in our repair shop). But you get a 10-bladed aperture with that. And yes, I’m aware a lot of you won’t consider a manually focusing lens. But some of you will.

MTF TESTS

For this section, please remember that the wide-open MTF is the average of 10 copies unless otherwise noted, taken at the widest aperture. As we’ll show later, MTF will be higher with smaller apertures, so the f/1.2 and f/0.95 lenses are at a disadvantage wide open, while the f/2.8 Macro-Elmarit has a big advantage.

So let’s change the order a bit and start from the smallest aperture to the largest; that should make scrolling up and down easier.

Panasonic Leica DG Macro Elmarit 45mm f/2.8 (8 copies only)

Well, just to get it out of the way, this isn’t a very good performance. MTF is fairly even with only a moderate amount of sagittal – tangential separation, but it’s low. It’s extraordinarily low given this is a f/2.8 lens, and particularly low at the higher frequencies which are where the fine detail lives. That being said, some macro lenses do perform better close up than at distant focus, which is where this test is done. But if you aren’t looking for a macro lens, there’s not much on the MTF chart to recommend this one.

Olympus M. Zuiko Digital 45mm f/1.8

This is a pretty impressive performance at a f/1.8 aperture; in fact, it’s far better at f/1.8 than the Macro was at f/2.8. There’s a bit of sag-tan separation and performance does drop off away from the center, but this is pretty good.

Panasonic Lumix 42.5mm f1.7 ASPH OIS (6 copies)

The smaller Panasonic also does very well. It has a little bit more separation of sagittal and tangential out at the edges of the image but otherwise is nearly identical to the f/1.8 Olympus as far as sharpness goes. I won’t hair-split the slightly wider aperture because m4/3 manufacturers run a bit fast and loose with f/# and I honestly can’t say for sure this is really wider than the f/1.8.

Olympus M. Zuiko Digital 45mm f/1.2 PRO

Remember you’re at much wider aperture now, so we expect a fall-off in MTF compared to the lenses above. The Digital Pro looks quite good for an ultra-wide aperture. It has a good resolution that it maintains across the field and a mild-to-moderate amount of sagittal-tangential separation.

Panasonic Leica DG Noctitron 42.5mm f1.2 ASPH Power OIS

The Panasonic f/1.2 is very similar to the Olympus. The MTF is a tiny bit higher in the center but lower at the edges, but probably not enough of either that you’d notice a difference in an image.

Voigtlander 42.5mm f/0.95 Aspherical

As expected, when we get to mega ultra-wide apertures, resolution falls off. This is a pretty decent performance for a f/0.95 lens, though.

Stop-Down Tests

These are not 10-copy tests; we picked one average copy of the lenses to do stop-downs on.

Voigtlander at f/1.4

This was the first easily measurable stop-down for the Voigt, just to get an idea if it would be similar to the f/1.2 lenses once it was stopped down a bit. The answer is yes, it is. In the middle 1/2 of the image, it’s actually sharper now than any of the others, although it fades quite a bit in the outer 1/3 of the image. Still, it’s nice to know you can get very high resolution in the center if the aperture is stopped down just a bit.

All Lenses at f/2.8

We’ll make side-by-side comparisons of each lens at wide open and f/2.8. Again, these are single copies so that the wide open MTF will be a little different than the averages above.

Olympus M. Zuiko Digital 45mm f/1.8

If your question is ‘how did it get sharper in the center, but softer in the edges when stopped down,’ well, the answer is it didn’t. This can happen when a lens has a lot of field curvature. We’re focused in the center for the MTF test, the field curvature (we’ll show that in a minute) is significant enough that the edges remain out of focus despite being stopped down. I explain in more detail in an addendum, if you’re interested.

Panasonic Lumix 42.5mm f1.7 ASPH OIS

Primarily a similar improvement in the center, but since this lens doesn’t have field curvature, a more predictable result away from the axis.

Olympus M. Zuiko Digital 45mm f/1.2 PRO

Really impressive f/2.8 performance here, and better at f/2.8 than either of the less expensive lenses.

Panasonic Leica DG Noctitron 42.5mm f1.2 ASPH Power OIS

Another impressive improvement, although not quite as sharp in the center as the Olympus. On the other hand, it holds sharpness better in the middle of the field before fading at the edges.

Voigtlander 42.5mm f/0.95 Aspherical

With its ultra-wide aperture, I’m impressed that the Voigtlander can sharpen up comparably to the two f/1.2 lenses. In the outer half of the image, it’s not quite as good, but I think within ‘sample variation’ of the other two.

Field Curvature (MTF vs. Field vs. Focus)

For those of you who don’t read our technical articles very often, the field curvature may be the most useful thing we give you. This is NOT distortion; rather it’s how the plane of best focus curves. The tangential and sagittal fields often curve differently; when they do you know, there will be astigmatism in those areas.

There’s other information you can get from field curvatures and if you’re interested here are some background articles: Fun with Fields of Focus 1, Fun with Fields of Focus II, Field Curvature and Stopping Down. I should mention these are done at f/5.6 because that gives a nice, clear picture of the field. Stopping down (or opening up) doesn’t change the curvature significantly. Keep in mind that when you’re shooting at a wider aperture, though, the field is much narrower.

Also, the fields shown are a single copy, so you’ll see some field tilt. That’s relatively common in these lenses.

Panasonic Leica DG Macro Elmarit 45mm f/2.8 (8 copies)

You’ll notice this field is both tilted and the tangential curve decentered. This is pretty typical for the copies of this lens we tested. It’s theoretically quite flat, which a macro lens should be, but in reality, is generally going to be a bit tilted.

Olympus M. Zuiko Digital 45mm f/1.8

The little Olympus has quite significant U-shaped curvature, and in a photo, you should quickly notice that it’s not going to get everything from side to side in focus, even stopped down. This explains the MTF fall-off at the edges we saw above, even stopped down. Note also that the tangential field curves more than the sagittal, so you’ll always have astigmatism (red area in the difference graph) out at the edges, again, even stopped down.

Panasonic Lumix 42.5mm f1.7 ASPH OIS (6 copies)

Now, how about this one for a nice, flat field with little astigmatism? I’ll take it. But you can notice there is a little curve in the sagittal field. Wide open with a narrower depth of field, this does give a little astigmatism, which you can see in the wide-open MTF graphs above. Still, very nice.

Olympus M. Zuiko Digital 45mm f/1.2 PRO (8 copies)

The Olympus Pro is also fairly flat. You can see a bit of inverted “U” in the tangential curvature, and some mild resultant edge astigmatism.

Panasonic Leica DG Noctitron 42.5mm f1.2 ASPH Power OIS

Another slightly tilted and decentered tangential field, but again the field is fairly flat. Notice how in this copy the field tilt results in increased astigmatism on one side of the lens. You might notice this one if you pixel peeped a bit.

Voigtlander 42.5mm f/0.95 Aspherical

The Voigtlander again surprises me. I didn’t expect so flat of a field in such a wider aperture lens.

Copy-to-Copy Variation

Micro 4/3 lenses, in general, have a lot of sample variation. Why this is I can’t say. What I can say is it’s not the fabled “they need to up their QA check” that people imagine happens. Micro 4/3 lenses, as a rule, don’t have any compensating adjustable elements, so what you get at the end of the assembly line is what you get unless something is broken inside.

Because we didn’t have 10 copies of each lens, I was hesitant to use our usual variance numbers and graphs; that system depends on having 10 copies. What it did show, in general terms, was that the Panasonic 45mm f/2.8 Macro has what we consider unacceptable amounts of variance, the other two Panasonics were fairly high. The two Olympus lenses and the Voigtlander were all in the ‘usual’ range for primes.

But I’ll let you judge for yourselves. For each lens below I’ve put thumbnails of our Full Frame MTF map for the first eight copies tested. Remember, with m4/3 you can’t make manufacturer generalizations. One brand looks better at these focal lengths. It is NOT any predictor of how any of the brands will look with lenses of different focal lengths.

Panasonic Leica DG Macro Elmarit 45mm f/2.8 (8 copies)

As I said above, this amount of variance is not OK, but a good copy is pretty good.

Olympus M. Zuiko Digital 45mm f/1.8

There’s obvious variance here, but if you are shooting centered subjects, these would all be good. It’s a good example of a lot of lab variation that probably doesn’t affect pictures very much.

Panasonic Lumix 42.5mm f1.7 ASPH OIS (6 copies)

Yes, you can tell the difference between the better two and the others.

Olympus M. Zuiko Digital 45mm f/1.2 PRO (8 copies)

This one is pretty danged impressive for a f/1.2. The center of sharpness is in the center of the lens pretty much in every copy. A couple aren’t quite as sharp as the others, but that’s the nature of the beast.

Panasonic Leica DG Noctitron 42.5mm f1.2 ASPH Power OIS

Here’s a great example of why we have internet arguments. Two of these eight lenses are better than any of the Olympus Pros; three are worse. Reviewers or photographers who compared one copy of each are probably going to draw different conclusions.

Voigtlander 42.5mm f/0.95 Aspherical

Lower resolution wide open because this is f/0.95. But for this ultra-wide aperture, I have to say this is very reasonable.

So What Did We Learn Today?

Well, I learned that no matter how much I dislike Voigtlander, the company, I sure do admire Voigtlander lenses. The performance of the 42.5mm f/0.95 is spectacular. Of course, you also have to be OK with manually focusing, but still, that’s a lens worth considering.

If you want a f/1.2 lens in this focal length, well, let’s face it, it will probably come down to whether you want Power OIS or not. If you don’t, the Olympus 42.5mm f/1.2 PRO is slightly less expensive and has less variation. But a good copy of the Panasonic 42.5 Noctitron is excellent.

Both of the less expensive lenses in this range deliver a lot of bang for the buck. They have very good image quality, and if you don’t need the widest aperture, they give you a lot for 1/3 the price of the wide-aperture duo. If I were most interested in this focal length for portraits, I’d probably want the Zuiko Digital f/1.8; that field curvature will help throw everything other than the subject out of focus. For landscapes or architectural shooting, the flat field of the Panasonic f/1.7 would be an advantage.

Roger Cicala and Aaron Closz

Lensrentals.com

March 2018

Addendum: Field Curvature and MTF

When we looked at the stop-down curves for the Olympus 45mm f1.8, I mentioned that off-axis the MTF didn’t raise because the field curvature was so significant. This is a discussion of why that is so, even stopped down.

First, let’s look at the field curvature of the lens at 20 lp/mm frequency (the orange lines in the MTF graphs above). Below is the sagittal and then tangential field curvatures. For each, I’ve put a black line across the graph at the center focus position.

If you wanted to mentally read the MTF as you follow the line from center to edge, you’d get curves very similar to the MTF measure above for f/2.8. You should notice that the sagittal field isn’t as curved as the tangential, and therefore the sagittal MTF is higher. But I’ll do that for you below (this is only going to 8mm away from the center, but that’s plenty to make the point).

You might ask the very pertinent question, “The MTF is higher away from the center at a different focus position than best center focus. What if I focus out on the edges”? Well, you’ll get a much sharper image. The lens is very sharp away from the center; it’s just that the sharp area is in a different focus plane. But we can graph the best MTF for each point and get something very different; a graph that shows how sharp the lens could be if you focus off-axis. We call this the BIF graph, for Best Individual Focus.

The BIF isn’t real, though, it’s theoretical. If you focused at that point 5mm away from the center, your image at that point would be very sharp, but then the center of the image would be very soft. So with a lens like this, you can get a sharp portrait, for example, anywhere in the field of view. But you can’t make the image sharp from one side to the other, even stopped down.

223 Comments

Lee · March 7, 2018
I dunno man, my Schneider PC-TS 90mm f/4.5 Makro-Symmar HM is pretty good for such a long name 😀
1. Roger Cicala · March 7, 2018
  They stopped making it, didn’t they? I blame the name length 🙂
  
  Actually, I loved that lens and kept waiting for it to explode in popularity. Which is why Lensrentals stopped letting me decide what to purchase.
Lee · March 7, 2018
I dunno man, my Schneider PC-TS 90mm f/4.5 Makro-Symmar HM is pretty good for such a long name :D
1. Roger Cicala · March 7, 2018
  They stopped making it, didn't they? I blame the name length :-)
  Actually, I loved that lens and kept waiting for it to explode in popularity. Which is why Lensrentals stopped letting me decide what to purchase.
YS · March 7, 2018
Thanks for the write-up. Good to know most of the ~45mm lenses are all good. Also ” For landscapes or architectural shooting, the flat field of the Olympus f/1.7 would be an advantage.” Surely you mean the Panasonic?
1. Roger Cicala · March 7, 2018
  I do and I fixed that. Thank you!!
YS · March 7, 2018
Thanks for the write-up. Good to know most of the ~45mm lenses are all good. Also " For landscapes or architectural shooting, the flat field of the Olympus f/1.7 would be an advantage." Surely you mean the Panasonic?
Mike Aubrey · March 7, 2018
It may be worth pointing readers to this article on macros, just so that we all see data on the difference between distance focus and close focusing MTF: https://www.lensrentals.com/blog/2016/08/more-photogeekery-finite-conjugate-mtf-bench-tests-for-macro-lenses/

Anyway, thanks as always. The best individual focus for the Olympus is exactly as I have experienced…and also the curvature…which can hurt the out of focus blur since it curves away from the sensor toward the distance in the outer frame.
1. Mike Aubrey · March 8, 2018
  But from Brandon’s comment about a positive vs. negative numbers. The Olympus’ curvature goes toward the lens from infinity?
  
  I’m going to have to check this on my copy…
  1. Brandon Dube · March 8, 2018
    Ah, christ. When I say “towards the lens” I mean the focal plane (where you stick your sensor) is closer to the rear element. This means the object that is in focus is at a more infinite distance (i.e., beyond infinity).
    
    I would use the language object plane to refer to where you would put a landscape or someone’s face or whatever.
    1. Mike Aubrey · March 8, 2018
      Ah okay. That’s what I had thought originally from the post. It isn’t you, it’s me. I keep getting myself turned around! 😉
      1. Hank Roest · March 9, 2018
        You simply have to focus on getting your subject to stand, or sit, beyond infinity. 😉
        
        Olaf Reinhard Weyer · March 31, 2018
        Beyond infinity… quantum mechanics can be mind boggling 😀
Mike Aubrey · March 7, 2018
It may be worth pointing readers to this article on macros, just so that we all see data on the difference between distance focus and close focusing MTF: https://www.lensrentals.com...
Anyway, thanks as always. The best individual focus for the Olympus is exactly as I have experienced...and also the curvature...which can hurt the out of focus blur since it curves away from the sensor toward the distance in the outer frame.
1. Mike Aubrey · March 8, 2018
  But from Brandon's comment about a positive vs. negative numbers. The Olympus' curvature goes toward the lens from infinity?
  I'm going to have to check this on my copy...
  1. Brandon Dube · March 8, 2018
    Ah, christ. When I say "towards the lens" I mean the focal plane (where you stick your sensor) is closer to the rear element. This means the object that is in focus is at a more infinite distance (i.e., beyond infinity).
    I would use the language object plane to refer to where you would put a landscape or someone's face or whatever.
    1. Mike Aubrey · March 8, 2018
      Ah okay. That's what I had thought originally from the post. It isn't you, it's me. I keep getting myself turned around! ;)
      1. Hank Roest · March 9, 2018
        You simply have to focus on getting your subject to stand, or sit, beyond infinity. ;-)
        Olaf Reinhard Weyer · March 31, 2018
        Beyond infinity... quantum mechanics can be mind boggling :D
Ilya Zakharevich · March 7, 2018
Roger, thanks for this interesting writeup?—?as usual!

However, I do not think that your addendum (which is very good standalone) would actually address the question you claim it addresses: why widely-open can be better than closed-down.

Essentially, you want to explain why MTF is 0.3 widely open vs. 0.05 closed to ƒ/2.8 (at the edge of the frame). Your explanation, essentially, goes like this: “In fact, this part of the image is out-of-focus w.r.t. to the curved surface of best focus”. Right?

However, closing down IMPROV?S out-of-focus areas?—?so if your explanation did hit the main contributor into softness, the effect would be exactly OPPOSITE to what is observed.

CONCLUSION: it is something else (in ADDITION to field curvature) which makes MTF worse when closed down…
1. Roger Cicala · March 8, 2018
  Ilya,
  
  You are correct, of course, but I was trying to not go there (and failed obviously). I explained why the MTF curve doesn’t improve, but not why it would deteriorate in places.
  
  The ‘deteriorate’ part probably has more to do with a bit of focus shift, either classic as we talk about on the camera, or bench focus shift (focus is done by detecting highest point spread function at center). With a flat field, it’s no matter, but with a wicked-steep curve it can shifts the MTF readings, particularly at higher frequencies where the field narrows, to an area where the reading is actually lower off center. I think that’s the more accurate explanation.
  
  It gets even more perplexing with the “W” field lenses, where we sometimes see the center and edges sharpen, and the ‘halfway’ area not improving.
  
  What I think would be better is showing the MTF v F v Focus at different apertures, one could then see the sharpening happening even though it’s not reflected in the MTF curve. But I fear my blog posts are rapidly becoming far too long.
  
  I probably
  1. Impulse_Vigil · March 8, 2018
    Was that a funny at the end of the comment? Keep the long blog posts coming IMO, there’s knowledge here that’s not easily gained anywhere else.
    1. Roger Cicala · March 8, 2018
      Really wasn’t meant to be funny at all; I agonize over the length of these already. I’m quite convinced the major purpose of my posts these days is to fight insomnia for photographers everywhere.
      
      But I HAVE to leave stuff out. If I was reporting complete data on just one of these lenses for industry it would run 30 pages, a couple of dozen excel spreadsheets, and 40 graphs – and I’m trying to compress 6 of them into one blog post.
      1. Impulse_Vigil · March 8, 2018
        I meant the line that seemingly got cut off (re.: funny), after the comment about length it reads “I probably” and then nothing! The effort is appreciated either way.
        
        Roger Cicala · March 8, 2018
        Ooops, that was just random Rogerness so I removed it. Usually this is some evidence of my multitasking inability. Probably I was typing the email while talking on the phone or something.
      2. Joe · March 14, 2018
        Please keep the lengthy posts and heavy data coming! There’s too little serious work out there in the three-bullet-points per slide PowerPoint world we’ve devolved into. Your detailed, data-driven approach is something that we can look forward to, and trust, compared to the buy-it-now click-bait fluff that we too often see elsewhere. (I expect that my thoughts are probably mainstream, at least for people who studied some physics back at MIT a long long time ago in a city far away . )
        
        I’ve tried a number of these M43 lenses on a Pen-F and an E-M5 II and my preference has become the Panasonic 42.5/1.7 The sharpness is excellent across the frame, at least in my copy, and the lens is small, light, and affordable.
        
        FWIW I recall from your bio that you also shoot a Pentax K-1. I tried to rent a Pentax 200/2.8 telephoto lens from LR to test on full-frame prior to possible purchase, but unfortunately you don’t list that model. Turns out that it’s an excellent full-frame lens on the K-1, sharp from corner to corner with no perceptible vignetting, at normal outdoor apertures. Nice metal build quality as well. Is the SDM focus mechanism that lens’s downfall?
  2. Ilya Zakharevich · March 8, 2018
    Roger, Brandon:
    
    I think what you wanted to say is, basically: the central part of the lens has very different field curvature than the outer part of the lens.
    
    Above, “central” means “near optical axis”.
    
    This is very familiar to me from my attempts to design a symmetric triplet in “exact optics”. This effect (it may even have a name: spheri-field-curvature? 😉 is what everybody misses when they claim that curved sensors are better than sliced bread!
    
    Anyway, without 2D maps of MTF (as colormaps above) taken separately: wide open and closed down, it may be hard to qualify.
    1. Roger Cicala · March 8, 2018
      I actually followed that, so obviously my coffee is kicking in. Or my reading. Now I have a question for you: would we also need maps at various frequencies, as well as apertures? (We have them, I’m very curious because I’m considering making a ‘slider’ app that lets us show the field at different apertures and frequencies and I’m wondering if we don’t need both options.
    2. Brandon Dube · March 8, 2018
      The (apparent) field curvature will only change when the aperture is changed because of oblique spherical aberration. On-axis’ best focus position will move as a function of aperture due to focus shift.
2. Brandon Dube · March 8, 2018
  Field curvature + focus shift. The focus is shifted away from the paraxial plane wide open on-axis. This focus shift is maybe 80 microns in magnitude (but obviously highly variable between different lens models, spectral content of light, etc…) while the field curvature may be, say, 300 microns. With the example of the 45/1.8, the field curvature of both S and T shifts in the positive direction, i.e. focus beyond infinity. If the focus shift is in the same direction, then as you close the aperture the on-axis focus position moves in the negative direction, opposite to the field curvature, and the image quality for a flat object will become very bad at the edges. This is what happens with that lens in particular.
  
  The focus shift, in the monochromatic case, is related to the relative balance of the various orders of spherical aberration. As you stop down, the third order spherical reduces by aperture^3, the fifth order aperture^5, and so on. Obviously this means the higher order terms rapidly vanish, and their balance, i.e. the ratio of the coefficient of any two orders to each other, changes as the aperture is closed.
  
  Since the focus shift from paraxial is associated with the balance of them, and that balance changes as a function f aperture, the focus shift from paraxial varies with aperture, and this is what we understand colloquially as focus shift.
  
  In principle we can derive the aberrations from MTF data, with more time for software and a very powerful computer we may begin doing that in the soonish future.
  1. Ilya Zakharevich · March 8, 2018
    > In principle we can derive the aberrations from MTF data, with more time
    for software and a very powerful computer we may begin doing that in
    the soonish future.
    
    Do not you need to know the complex (meaning ??) transfer function, not only its absolute value, to do this? I think one may be reconstructed from the other, but do not remember how numerically stable the result would be. (This might be similar to the question “deconvolve when all you know is MTF”.) And since you know |?-MTF| only in 10 frequencies (5 tangential and 5 saggital), this may be quite a trick to achieve!
    1. Roger Cicala · March 8, 2018
      Ilya, I know you and Brandon are about to start talking math over my head, but we actually get data up to 300 lp/mm in 5 lp/mm increments; it’s just not graphed. I may screw this up (and Brandon will happily correct me in a bit, I’m sure), but I think it requires that kind of data in the MTFvFvFocus format so we see defocus areas, and at several apertures.
    2. Brandon Dube · March 8, 2018
      You don’t need the Optical Transfer Function (OTF), the MTF will do. You just have to extend an image-based wavefront sensing (WFS) program to use MTF data instead of PSFs. It is not a directly mathematical process, because even between the aberrations and PSFs there is a loss of Fourier phase. Going to the MTF loses you Fourier phase again, and when you just have Tan,Sag it becomes an azimuthally limited subset of that.
      
      If you write a good enough WFS algorithm, it can be done.
  2. Athanasius Kirchner · March 9, 2018
    That’s a brilliant explanation, thanks.
    The worst case scenario for corner sharpness, then, is obviously when the plane of focus is very curved, and there’s focus shift in the same direction.
    How do images look when both sagittal and tangential planes of focus curve in opposite directions? I’d expect better resolution than in the case of the Olympus 45mm f/1.8, or is a good sagittal performance offset by a terrible tangential one (or vice-versa)?
    1. Brandon Dube · March 10, 2018
      If there’s a lot of astigmatism (T and S curve in opposite direction or one curves and one does not) then the image has highly directional defocus. If you watch quite a lot of movies – especially from the 80s and 90s using anamorphic lenses – the very top and bottom of the frame is really blurry. That’s what astigmatism looks like in a picture when taken to an extreme.
      
      I can simulate images of astigmatism from a reference frame, but not with the proper dependence over the field of view (which is a significant part of the “look”). Simulating with the dependence of the FoV efficiently is a highly complicated algorithm I haven’t implemented yet.
Ilya Zakharevich · March 7, 2018
Roger, thanks for this interesting writeup — as usual!
However, I do not think that your addendum (which is very good standalone) would actually address the question you claim it addresses: why widely-open can be better than closed-down.
Essentially, you want to explain why MTF is 0.3 widely open vs. 0.05 closed to ƒ/2.8 (at the edge of the frame). Your explanation, essentially, goes like this: “In fact, this part of the image is out-of-focus w.r.t. to the curved surface of best focus”. Right?
However, closing down IMPROVΕS out-of-focus areas — so if your explanation did hit the main contributor into softness, the effect would be exactly OPPOSITE to what is observed.
CONCLUSION: it is something else (in ADDITION to field curvature) which makes MTF worse when closed down…
1. Roger Cicala · March 8, 2018
  Ilya,
  You are correct, of course, but I was trying to not go there (and failed obviously). I explained why the MTF curve doesn't improve, but not why it would deteriorate in places.
  The 'deteriorate' part probably has more to do with a bit of focus shift, either classic as we talk about on the camera, or bench focus shift (focus is done by detecting highest point spread function at center). With a flat field, it's no matter, but with a wicked-steep curve it can shifts the MTF readings, particularly at higher frequencies where the field narrows, to an area where the reading is actually lower off center. I think that's the more accurate explanation.
  It gets even more perplexing with the "W" field lenses, where we sometimes see the center and edges sharpen, and the 'halfway' area not improving.
  What I think would be better is showing the MTF v F v Focus at different apertures, one could then see the sharpening happening even though it's not reflected in the MTF curve. But I fear my blog posts are rapidly becoming far too long.
  1. Impulse_Vigil · March 8, 2018
    Was that a funny at the end of the comment? Keep the long blog posts coming IMO, there's knowledge here that's not easily gained anywhere else.
    1. Roger Cicala · March 8, 2018
      Really wasn't meant to be funny at all; I agonize over the length of these already. I'm quite convinced the major purpose of my posts these days is to fight insomnia for photographers everywhere.
      But I HAVE to leave stuff out. If I was reporting complete data on just one of these lenses for industry it would run 30 pages, a couple of dozen excel spreadsheets, and 40 graphs - and I'm trying to compress 6 of them into one blog post.
      1. Impulse_Vigil · March 8, 2018
        I meant the line that seemingly got cut off (re.: funny), after the comment about length it reads "I probably" and then nothing! The effort is appreciated either way.
      2. Joe · March 14, 2018
        Please keep the lengthy posts and heavy data coming! There's too little serious work out there in the three-bullet-points per slide PowerPoint world we've devolved into. Your detailed, data-driven approach is something that we can look forward to, and trust, compared to the buy-it-now click-bait fluff that we too often see elsewhere. (I expect that my thoughts are probably mainstream, at least for people who studied some physics back at MIT a long long time ago in a city far away <g>. )
        I've tried a number of these M43 lenses on a Pen-F and an E-M5 II and my preference has become the Panasonic 42.5/1.7 The sharpness is excellent across the frame, at least in my copy, and the lens is small, light, and affordable.
        FWIW I recall from your bio that you also shoot a Pentax K-1. I tried to rent a Pentax 200/2.8 telephoto lens from LR to test on full-frame prior to possible purchase, but unfortunately you don't list that model. Turns out that it's an excellent full-frame lens on the K-1, sharp from corner to corner with no perceptible vignetting, at normal outdoor apertures. Nice metal build quality as well. Is the SDM focus mechanism that lens's downfall?
  2. Ilya Zakharevich · March 8, 2018
    Roger, Brandon:
    I think what you wanted to say is, basically: the central part of the lens has very different field curvature than the outer part of the lens.
    Above, “central” means “near optical axis”.
    This is very familiar to me from my attempts to design a symmetric triplet in “exact optics”. This effect (it may even have a name: spheri-field-curvature? ;-) is what everybody misses when they claim that curved sensors are better than sliced bread!
    Anyway, without 2D maps of MTF (as colormaps above) taken separately: wide open and closed down, it may be hard to qualify.
    1. Roger Cicala · March 8, 2018
      I actually followed that, so obviously my coffee is kicking in. Or my reading. Now I have a question for you: would we also need maps at various frequencies, as well as apertures? (We have them, I'm very curious because I'm considering making a 'slider' app that lets us show the field at different apertures and frequencies and I'm wondering if we don't need both options.
    2. Brandon Dube · March 8, 2018
      The (apparent) field curvature will only change when the aperture is changed because of oblique spherical aberration. On-axis' best focus position will move as a function of aperture due to focus shift.
2. Brandon Dube · March 8, 2018
  Field curvature + focus shift. The focus is shifted away from the paraxial plane wide open on-axis. This focus shift is maybe 80 microns in magnitude (but obviously highly variable between different lens models, spectral content of light, etc...) while the field curvature may be, say, 300 microns. With the example of the 45/1.8, the field curvature of both S and T shifts in the positive direction, i.e. focus beyond infinity. If the focus shift is in the same direction, then as you close the aperture the on-axis focus position moves in the negative direction, opposite to the field curvature, and the image quality for a flat object will become very bad at the edges. This is what happens with that lens in particular.
  The focus shift, in the monochromatic case, is related to the relative balance of the various orders of spherical aberration. As you stop down, the third order spherical reduces by aperture^3, the fifth order aperture^5, and so on. Obviously this means the higher order terms rapidly vanish, and their balance, i.e. the ratio of the coefficient of any two orders to each other, changes as the aperture is closed.
  Since the focus shift from paraxial is associated with the balance of them, and that balance changes as a function f aperture, the focus shift from paraxial varies with aperture, and this is what we understand colloquially as focus shift.
  In principle we can derive the aberrations from MTF data, with more time for software and a very powerful computer we may begin doing that in the soonish future.
  1. Ilya Zakharevich · March 8, 2018
    > In principle we can derive the aberrations from MTF data, with more time
    for software and a very powerful computer we may begin doing that in
    the soonish future.
    Do not you need to know the complex (meaning ∈ℂ) transfer function, not only its absolute value, to do this? I think one may be reconstructed from the other, but do not remember how numerically stable the result would be. (This might be similar to the question “deconvolve when all you know is MTF”.) And since you know |ℂ-MTF| only in 10 frequencies (5 tangential and 5 saggital), this may be quite a trick to achieve!
    1. Roger Cicala · March 8, 2018
      Ilya, I know you and Brandon are about to start talking math over my head, but we actually get data up to 300 lp/mm in 5 lp/mm increments; it's just not graphed. I may screw this up (and Brandon will happily correct me in a bit, I'm sure), but I think it requires that kind of data in the MTFvFvFocus format so we see defocus areas, and at several apertures.
    2. Brandon Dube · March 8, 2018
      You don't need the Optical Transfer Function (OTF), the MTF will do. You just have to extend an image-based wavefront sensing (WFS) program to use MTF data instead of PSFs. It is not a directly mathematical process, because even between the aberrations and PSFs there is a loss of Fourier phase. Going to the MTF loses you Fourier phase again, and when you just have Tan,Sag it becomes an azimuthally limited subset of that.
      If you write a good enough WFS algorithm, it can be done.
  2. Athanasius Kirchner · March 9, 2018
    That's a brilliant explanation, thanks.
    The worst case scenario for corner sharpness, then, is obviously when the plane of focus is very curved, and there's focus shift in the same direction.
    How do images look when both sagittal and tangential planes of focus curve in opposite directions? I'd expect better resolution than in the case of the Olympus 45mm f/1.8, or is a good sagittal performance offset by a terrible tangential one (or vice-versa)?
    1. Brandon Dube · March 10, 2018
      If there's a lot of astigmatism (T and S curve in opposite direction or one curves and one does not) then the image has highly directional defocus. If you watch quite a lot of movies - especially from the 80s and 90s using anamorphic lenses - the very top and bottom of the frame is really blurry. That's what astigmatism looks like in a picture when taken to an extreme.
      I can simulate images of astigmatism from a reference frame, but not with the proper dependence over the field of view (which is a significant part of the "look"). Simulating with the dependence of the FoV efficiently is a highly complicated algorithm I haven't implemented yet.
Dariano Bisotnik · March 7, 2018
Roger, with compelling data like this you are the party-pooper for forumtographers.
One can simply link your article in reply to their arguments of “this lens is best… blah blah” and suddenly there could be just “crickets” 🙂
Thank you for the great work you do.
Dariano Bisotnik · March 7, 2018
Roger, with compelling data like this you are the party-pooper for forumtographers.
One can simply link your article in reply to their arguments of "this lens is best... blah blah" and suddenly there could be just "crickets" :-)
Thank you for the great work you do.
Deanaaargh · March 8, 2018
Roger,
This made for excellent reading, as usual. I was caught however by your phrase ‘as a rule’ m43 lenses do not feature adjustable elements. I doubt that this is something that is covered by the m43 coalition specification. I wonder whether it is because Panasonic/Olympus have relatively small repair operations and are not staffed up to provide re-centering services etc.
However, I am curious if you would feel comfortable estimating the proportion of new lenses produced today that do not allow for post assembly adjustment. I have some sense based upon the teardowns you have posted on this (excellent) blog but I would appreciate your considered opinion.
1. Roger Cicala · March 8, 2018
  Deanaaargh,
  
  I don’t recall ever seeing an adjustable element in an m4/3 lens, and the manufacturer’s repair centers in the US don’t seem able to do optical adjustments. BUT we don’t open up nearly as many m4/3 lenses as we do others so my ‘I’ve never seen’ doesn’t mean as much as with some lens types.
  
  I suspect some of the physically larger lenses (200 f/2.8, 300 f/4 IS, or some of the big zooms) might have some. We argued amongst ourselves if one of the medium telezooms had a centering front element, on the basis that we screwed a couple up cleaning out dust and had to spend some time getting them back right; but we decided it was more about us not marking position properly then it actually being a centering element.
  1. Paul Stuart · March 16, 2018
    does this mean you will do a telephoto lens test 300mm f4 200mm 2.8 with respective t/c also?
    1. Roger Cicala · March 16, 2018
      Nope, it most definitely does not mean that. Partly because the 200 f2.8 is electromagnetic focus. We get results when we tested it, but I don’t trust them. And I’m not investing the money it takes to make an electronic m4/3 mount any time soon.
      1. Paul Stuart · March 16, 2018
        How or why is electromagnetic focus so detremental to the tests .
        Could share the results when you have tested it
        
        Someone · March 16, 2018
        Because these lenses must be continuously connected to camera to keep focus. This, in turn, means the test bench needs a mount with electrical contacts connected to a camera.
        
        With traditional lenses, you can prefocus them to infinity, and they will keep focus without being connected to a camera for a duration of a test.
        
        Roger Cicala · March 18, 2018
        Paul, I wouldn’t feel comfortable. I get results, they look OK, but to get them we’re focusing at a different focus plane than normal because the focus isn’t at infinity. I don’t know how much that makes the results worse than reality, or if it does at all. So basically I’d be putting out a scientific looking graph with “I have no idea what this means” on it. There’s way to much of that on the internet already.
    2. Roger Cicala · March 16, 2018
      Nope, it most definitely does not mean that. Partly because the 200 f2.8 is electromagnetic focus. We get results when we tested it, but I don't trust them. And I'm not investing the money it takes to make an electronic m4/3 mount any time soon.
      1. Paul Stuart · March 16, 2018
        How or why is electromagnetic focus so detremental to the tests .
        Could share the results when you have tested it
        Someone · March 16, 2018
        Because these lenses must be continuously connected to camera to keep focus. This, in turn, means the test bench needs a mount with electrical contacts connected to a camera.
        With traditional lenses, you can prefocus them to infinity, and they will keep focus without being connected to a camera for a duration of a test.
        Roger Cicala · March 18, 2018
        Paul, I wouldn't feel comfortable. I get results, they look OK, but to get them we're focusing at a different focus plane than normal because the focus isn't at infinity. I don't know how much that makes the results worse than reality, or if it does at all. So basically I'm putting out a scientific looking graph with "I have no idea what this means" on it. There's way to much of that on the internet already.
Deanaaargh · March 8, 2018
Roger,
This made for excellent reading, as usual. I was caught however by your phrase 'as a rule' m43 lenses do not feature adjustable elements. I doubt that this is something that is covered by the m43 coalition specification. I wonder whether it is because Panasonic/Olympus have relatively small repair operations and are not staffed up to provide re-centering services etc.
However, I am curious if you would feel comfortable estimating the proportion of new lenses produced today that do not allow for post assembly adjustment. I have some sense based upon the teardowns you have posted on this (excellent) blog but I would appreciate your considered opinion.
1. Roger Cicala · March 8, 2018
  Deanaaargh,
  I don't recall ever seeing an adjustable element in an m4/3 lens, and the manufacturer's repair centers in the US don't seem able to do optical adjustments. BUT we don't open up nearly as many m4/3 lenses as we do others so my 'I've never seen' doesn't mean as much as with some lens types.
  I suspect some of the physically larger lenses (200 f/2.8, 300 f/4 IS, or some of the big zooms) might have some. We argued amongst ourselves if one of the medium telezooms had a centering front element, on the basis that we screwed a couple up cleaning out dust and had to spend some time getting them back right; but we decided it was more about us not marking position properly then it actually being a centering element.
cpt kent · March 8, 2018
Hi,
I may have missed it in the article (or another) but are all the lenses that you test ‘straight out of the box’, or ones that have been rented? I just wonder if some lenses go out of tolerance quicker than others? (I’m specifically thinking that the Nocticron has been around for a while, and that it has OIS). Thoughts?
1. Roger Cicala · March 8, 2018
  They weren’t all out of the box, but all were pretty new, less than a year old. Plus we test them (not to this degree, of course) before and after each rental.
  1. cpt kent · March 8, 2018
    Thanks. It’s probably my imagination, but I’ve always shied away from OIS due to uncertainty about long term durability. Seems to me, the less moving parts, the more durable something should generally be…
    1. Roger Cicala · March 8, 2018
      We definitely see a slightly higher failure rate in OIS lenses, as you say more parts to break. When the OIS breaks it definitely screws up the optics, but it’s not a subtle thing like the lens gets softer, it’s ‘it was fine, now it looks awful’.
      1. Carleton Foxx · March 9, 2018
        If I may offer a bit of real-world observation, I’ve watched guys load the trucks at my local Fedex office and they regularly heave the lighter boxes 10 feet into the back of their trucks, so the fact that OIS elements work at all is a testament to their well-built-ness.
cpt kent · March 8, 2018
Hi,
I may have missed it in the article (or another) but are all the lenses that you test 'straight out of the box', or ones that have been rented? I just wonder if some lenses go out of tolerance quicker than others? (I'm specifically thinking that the Nocticron has been around for a while, and that it has OIS). Thoughts?
1. Roger Cicala · March 8, 2018
  They weren't all out of the box, but all were pretty new, less than a year old. Plus we test them (not to this degree, of course) before and after each rental.
  1. cpt kent · March 8, 2018
    Thanks. It's probably my imagination, but I've always shied away from OIS due to uncertainty about long term durability. Seems to me, the less moving parts, the more durable something should generally be...
    1. Roger Cicala · March 8, 2018
      We definitely see a slightly higher failure rate in OIS lenses, as you say more parts to break. When the OIS breaks it definitely screws up the optics, but it's not a subtle thing like the lens gets softer, it's 'it was fine, now it looks awful'.
      1. Carleton Foxx · March 9, 2018
        If I may offer a bit of real-world observation, I've watched guys load the trucks at my local Fedex office and they regularly heave the lighter boxes 10 feet into the back of their trucks, so the fact that OIS elements work at all is a testament to their well-built-ness.
David B · March 8, 2018
I’ve owned many of these lenses over the years and I agree with conclusion. Ultimately I settled on 45/1.8 I am happy with rendering and size, size makes it so easy to take when traveling https://uploads.disquscdn.com/images/1793777f9ba79be60373fc97ef4d82576bbcfa2804203542122a847c26d67769.jpg
1. Justin · March 12, 2018
  Nice bokeh!!! 🙂
David B · March 8, 2018
I’ve owned many of these lenses over the years and I agree with conclusion. Ultimately I settled on 45/1.8 I am happy with rendering and size, size makes it so easy to take when traveling https://uploads.disquscdn.c...
1. NoGodsNoBosses · March 12, 2018
  Nice bokeh!!! :)
J L Williams · March 8, 2018
I always enjoy these articles, but this time I even got a practical takeaway. I use the Olympus 45/1.8, and out of longtime habit (I learned on rangefinder cameras) I almost always focus at the center and then recompose; with most lenses it makes little difference. I can see now, though, that when using the 45 I’ll be better off composing first and then moving the camera’s AF point to where the main subject is, so it will adjust for the field curvature at that position. Right?
1. Impulse_Vigil · March 8, 2018
  It’s a good practice in general if you have the time or aren’t averse to using the touchscreen to set the focus point (which I find faster on any camera w/a joystick for it)…
J L Williams · March 8, 2018
I always enjoy these articles, but this time I even got a practical takeaway. I use the Olympus 45/1.8, and out of longtime habit (I learned on rangefinder cameras) I almost always focus at the center and then recompose; with most lenses it makes little difference. I can see now, though, that when using the 45 I'll be better off composing first and then moving the camera's AF point to where the main subject is, so it will adjust for the field curvature at that position. Right?
1. Impulse_Vigil · March 8, 2018
  It's a good practice in general if you have the time or aren't averse to using the touchscreen to set the focus point (which I find faster on any camera w/a joystick for it)...
David Bateman · March 8, 2018
Thank you Roger for an other M43rds review. I am glad to see your done playing with minivans.
I find it interesting that you recommend the Olympus f1.8 for portrait and the Panasonic for others. I would have thought the field curvature would hurt, as part of the background could come into focus, depending on subject to background distance. I think for close up flower shots the curvature would be benificial with the Olympus. But now that we know it helps for the final image plan.
I do wonder about the large variation in M43rds lenses. If this is due to not having adjustable elements to correct line errors and how the camera software handles this. If DFD and internal camera lens corrections can take this into account. Or if internal corrections act to average out the variations.
1. Roger Cicala · March 8, 2018
  David, in the stop down curves, both sides are the same copy taken at different apertures.
  1. David Bateman · March 8, 2018
    Thank you, that makes sense as it looks like the selected Panasonic was less than the average and the selected Olympus was at the average.
    So depending on copy they may have similar center and Panasonic will have better mid to edge sharpness. That is about what I thought.
    So can we expect a 15-17 test next?
    Down at the wide angles, every focal length is covered so maybe hard to draw the line. With 12, 14,15,16,17,18,19,20 lenses to select from. Maybe a 13mm or 11mm lens will be released in the future.
    Thank you for all the fun hard work.
  2. David Bateman · March 8, 2018
    Thank you, that makes sense as it looks like the selected Panasonic was less than the average and the selected Olympus was at the average.
    So depending on copy they may have similar center and Panasonic will have better mid to edge sharpness. That is about what I thought.
    So can we expect a 15-17 test next?
    Down at the wide angles, every focal length is covered so maybe hard to draw the line. With 12, 14,15,16,17,18,19,20 lenses to select from. Maybe a 13mm or 11mm lens will be released in the future.
    Thank you for all the fun hard work.
David Bateman · March 8, 2018
Thank you Roger for an other M43rds review. I am glad to see your done playing with minivans.
I find it interesting that you recommend the Olympus f1.8 for portrait and the Panasonic for others. I would have thought the field curvature would hurt, as part of the background could come into focus, depending on subject to background distance. I think for close up flower shots the curvature would be benificial with the Olympus. But now that we know it helps for the final image plan.
I do wonder about the large variation in M43rds lenses. If this is due to not having adjustable elements to correct line errors and how the camera software handles this. If DFD and internal camera lens corrections can take this into account. Or if internal corrections act to average out the variations.
Also a question about the stopped down test on a single copy. Is the right curve plot at max aperture also just for the single copy? As I noticed that the Panasonic f1.2 center now starts at 0.4, rather than 0.5 for the group average.
tom · March 8, 2018
Hello… Great work again and actually the best technical evaluation of optics around the Web…

In his publication i wonder whether the curves of the Olympus 45 1.2 and Pana 42.5 1.2 are differing in the charts at the beginning and when stopped down… my personal experience when using both: the latter seem to be more realistic because the olympus really is sharper already in the center and especially in the corners
tom · March 8, 2018
Hello... Great work again and actually the best technical evaluation of optics around the Web...
In his publication i wonder whether the curves of the Olympus 45 1.2 and Pana 42.5 1.2 for ' 1.2 ' are differing in the charts at the beginning and when stopped down... my personal experience when using both: the latter seem to be more realistic because the olympus really is sharper already in the center and especially in the corners
Impulse_Vigil · March 8, 2018
Very interesting tests and info! I hope there’s more M4/3 testing on the way, or at least a similar grouping of wides (even tho the focal length selection for the system is all over the place there)… Pretty please?

I’ve read a few reviews of the 45/1.8 & 42.5/1.7, and seen a few comparisons in the boards, and not once did I ever see anyone suggest there might be a field curvature issue or difference (depending on how you look at it); even tho it’s something that’s come up for other lenses (Olys in particular, like the 17/1.8 & 7-14).

That actually helps explain some of the differences and conclusions I’ve read regarding comparisons between those two. I *think* I’ve read the linked article regarding the subject, but I’ve might’ve missed this nonetheless, is there an easy-ish way to figure out if there’s some significant field curvature for a lens at home?

Worth pointing that the Pana 42.5/1.7 has OIS too *and* it focuses much closer. FWIW I owned both and kept the latter mostly because of this rather than any of the IQ/bokeh differences that get over analysed on the boards/reviews (w/o access to multiple samples etc).
1. Roger Cicala · March 8, 2018
  Very good point on the closer focusing capabilities on the Pana.
  1. Mike Whitten · March 8, 2018
    FWIW, the specs say the Pana f1.7 will close focus to 31cm, indeed much better than the Oly f1.8’s 50cm. That lump of a Voigtlander… 23cm. I’ve got the Oly f1.8 and the Voigtlander, which is big and heavy and as pointed out, manual. Sure does make nice images when one can live with the trade-offs.
  2. Mike Whitten · March 8, 2018
    FWIW, the specs say the Pana f1.7 will close focus to 31cm, indeed much better than the Oly f1.8's 50cm. That lump of a Voigtlander... 23cm. I've got the Oly f1.8 and the Voigtlander, which is big and heavy and as pointed out, manual. Sure does make nice images when one can live with the trade-offs.
2. Scintilla · March 9, 2018
  > is there an easy-ish way to figure out if there’s some significant field curvature for a lens at home?
  Based on what I’ve gotten from these articles so far:
  
  Shoot a chart or brick wall focused in the center, then shoot again focused in a corner.
  If there’s no difference, then no field curvature (unless it’s “M-shaped”).
  If all the corners got better, then it’s field curvature.
  If the corner of focus got better but the opposite corner got worse, then it’s field tilt (and possibly also field curvature).
  
  Did I get it right?
  1. Roger Cicala · March 16, 2018
    That will work. Here’s an easier one:
    1) put a rock or some other focus object in your yard.
    2) Carefully focus on the object and take a picture wide open.
    3) Put the picture in Photoshop and run a find edges filter, adjust to find just the sharpest edges. You’ll get a nice picture of the field curvature.
    
    For more detail, look at the examples in the last section of this: https://www.lensrentals.com/blog/2016/09/fun-with-field-of-focus-part-1/
    1. Scintilla · March 16, 2018
      Ah, the ever-versatile Rock Or Something!
      
      You know, I don’t know why I didn’t think to suggest doing something very similar with a sheet of graph paper on a table, considering I’ve done exactly that before, though obviously the field curvature at such close distances may not be the same as at the distances you’d get with your method. Thanks Roger!
Impulse_Vigil · March 8, 2018
Very interesting tests and info! I hope there's more M4/3 testing on the way, or at least a similar grouping of wides (even tho the focal length selection for the system is all over the place there)... Pretty please?
I've read a few reviews of the 45/1.8 & 42.5/1.7, and seen a few comparisons in the boards, and not once did I ever see anyone suggest there might be a field curvature issue or difference (depending on how you look at it); even tho it's something that's come up for other lenses (Olys in particular, like the 17/1.8 & 7-14).
That actually helps explain some of the differences and conclusions I've read regarding comparisons between those two. I *think* I've read the linked article regarding the subject, but I've might've missed this nonetheless, is there an easy-ish way to figure out if there's some significant field curvature for a lens at home?
Worth pointing that the Pana 42.5/1.7 has OIS too *and* it focuses much closer. FWIW I owned both and kept the latter mostly because of this rather than any of the IQ/bokeh differences that get over analysed on the boards/reviews (w/o access to multiple samples etc).
1. Scintilla · March 9, 2018
  > is there an easy-ish way to figure out if there's some significant field curvature for a lens at home?
  Based on what I've gotten from these articles so far:
  Shoot a chart or brick wall focused in the center, then shoot again focused in a corner.
  If there's no difference, then no field curvature (unless it's "M-shaped").
  If all the corners got better, then it's field curvature.
  If the corner of focus got better but the opposite corner got worse, then it's field tilt (and possibly also field curvature).
  Did I get it right?
  1. Roger Cicala · March 16, 2018
    That will work. Here's an easier one:
    1) put a rock or some other focus object in your yard.
    2) Carefully focus on the object and take a picture wide open.
    3) Put the picture in Photoshop and run a find edges filter, adjust to find just the sharpest edges. You'll get a nice picture of the field curvature.
    For more detail, look at the examples in the last section of this: https://www.lensrentals.com...
    1. Scintilla · March 16, 2018
      Ah, the ever-versatile Rock Or Something!
      <img src="https://taskandpurpose.com/...">
      You know, I don't know why I didn't think to suggest doing something very similar with a sheet of graph paper on a table, considering I've done exactly that before, though obviously the field curvature at such close distances may not be the same as at the distances you'd get with your method. Thanks Roger!
Ivar Brekke · March 8, 2018
Interesting article as always. There is a lot more characteristics of any lens than sharpness though, that most online testers do not consider, but is clearly seen when comparing lenses of similar focal length side by side. I would even argue that these other characteristics (example colour, dynamic range and bokeh) are more important and what often separates the premium lenses. I have been using m43 for years and after a while, you certainly see that some lenses shoot nicer pictures than others lenses with similar field of view in similar circumstances.
1. Ivar Brekke · March 8, 2018
  Regarding who designs, produces and brands lenses, with many actors in the process, is this not how most industries work today? like an iPhone with one designer in California, one producer of chipset, one of glass, one of camera module, and one who sets it all together, etc. Still it is clearly an iPhone. I think the premium brands is just a sign that these are designed and produced at a higher standard and with a consistent design, it beeing cosmetical or optical.
  1. Roger Cicala · March 8, 2018
    Ivar, it certainly is in industry in general. Many lens manufacturers do things nearly all or all in house (Canon, Sigma, Zeiss and Leica; probably Nikon). Some sublet all of their glass one place, electronics somewhere else, assembly somewhere else. And all phases in between.
    
    Premium means whatever you want it to, I think, depending on how you define it. But don’t overestimate how much brands check and know about there stuff. I make most of my modest living doing rechecks for industry; I still find it shocking how often I say, “So the supplier promised you a certain spec, you couldn’t check it, so you just believed them?”
    1. Ivar Brekke · March 9, 2018
      If I remember correctly Sigma has a m43 25mm f1.4 patent on a lens design that looks identical to the PL 25. If Sigma are the designer and producer of the lens, would I trust them to produce good quality? I think I would. Panasonic are also investing in moulding of aspherical lenses, so they probably produce some parts themselves, and also deliver elements to many other lens brands.
      
      Premium definition in the Oxford dictionary: An amount that is more than usual. If I could choose freely between the Olympus 45 f1.8 or the f1.2 I would always choose the f1.2 indicating that it is a premium lens. If I could choose freely between the Panasonic 12-60 vs the PL 12-60, I would always choose the PL, I think most people would, and I believe the quality on average would be some steps up.
      
      You could question that Panasonic use the Leica branding for their premium/pro lenses, but these company do seem to have a close collaboration. Many Panasonic cameras comes in both Panasonic and Leica branding. Also cameras like the Leica Q and SL line have a lot of resemblance with the Panasonic cameras, indicating that Leica are using a similar strategy and that the cooperation is a two way thing. Leica is a much stronger brand in the photo universe than Panasonic. why not benefit on it? It is a business after all.
      
      You could even argue that businesses that do everything inhouse at one point will have a harder time adapting to changes in technology and marketplace. And you can kind of see it in Canon and Nikons slow adaption of mirrorless.
      
      Still I can see that a lens could be considered premium if it is hand made according to years of tradition and perfection by some japanese optical master, much like a cuban cigar has a premium brand.
    2. Athanasius Kirchner · March 9, 2018
      Do Cosina do most of their lenses in-house as well? Seeing that they’re one of the better-known for-hire lens manufacturers, I’d have thought so.
Ivar Brekke · March 8, 2018
Interesting article as always. There is a lot more characteristics of any lens than sharpness though, that most online testers do not consider, but is clearly seen when comparing lenses of similar focal length side by side. I would even argue that these other characteristics (example colour, dynamic range and bokeh) are more important and what often separates the premium lenses. I have been using m43 for years and with experience, you certainly see that some lenses shoot nicer pictures than others lenses with similar field of view in similar circumstances.
1. Ivar Brekke · March 8, 2018
  Regarding who design, produce and brand lenses, with many actors in the process, is this not how most industries work today? like an iPhone with one designer in California, one producer of chipset, one of glass, one of camera module, and one who sets it all together, etc. Still it is clearly an iPhone. I think the premium brands is just a sign that these are designed and produced at a higher standard and with a consistent design, it beeing cosmetical or optical.
  1. Roger Cicala · March 8, 2018
    Ivar, it certainly is in industry in general. Many lens manufacturers do things nearly all or all in house (Canon, Sigma, Zeiss and Leica; probably Nikon). Some sublet all of their glass one place, electronics somewhere else, assembly somewhere else. And all phases in between.
    Premium means whatever you want it to, I think, depending on how you define it. But don't overestimate how much brands check and know about their stuff. I make most of my modest living doing rechecks for industry; I still find it shocking how often I say, "So the supplier promised you a certain spec, you couldn't check it, so you just believed them?"
    1. Ivar Brekke · March 9, 2018
      If I remember correctly Sigma has a m43 25mm f1.4 patent on a lens design that looks identical to the PL 25. If Sigma are the designer and producer of the lens, would I trust them to produce good quality? I think I would. Panasonic are also investing in moulding of aspherical lenses, so they probably produce some parts themselves, and also deliver elements to many other lens brands. The Quality Assurence probably consist of taking out a sample from the production line once now and then.
      Premium definition in the Cambridge dictionary: An amount that is more than usual. If I could choose freely between the Olympus 45 f1.8 or the f1.2 I would always choose the f1.2 indicating that it is a premium lens. If I could choose freely between the Panasonic 12-60 vs the PL 12-60, I would always choose the PL, I think most people would, and I believe the quality on average would be some steps up.
      You could question that Panasonic use the Leica branding for their premium/pro lenses, but these companies do seem to have a close collaboration. Many Panasonic cameras comes in both Panasonic and Leica branding. Also cameras like the Leica Q and SL line have a lot of resemblance with the Panasonic cameras, indicating that Leica are using a similar strategy and that the cooperation is a two way thing. Leica is a much stronger brand in the photo universe than Panasonic, why not benefit on it? It is a business after all.
      You could even argue that businesses that do everything inhouse at one point will have a harder time adapting to changes in technology and marketplace. And you can kind of see it in Canon and Nikons slow adaption of mirrorless.
      Still I can see that a lens could be considered premium if it is hand made according to years of tradition and perfection by some japanese optical master, much like a cuban cigar has a premium brand.
    2. Athanasius Kirchner · March 9, 2018
      Do Cosina do most of their lenses in-house as well? Seeing that they're one of the better-known for-hire lens manufacturers, I'd have thought so.
Samuel H · March 8, 2018
I have a doubt about your statement that “for portraits, I’d probably want the Zuiko Digital f/1.8; that field curvature will help throw everything other than the subject out of focus”.

I hope this means that the curve in these graphs means the sides are focused closer than the center is. If it means that the center is focusing closer than the sides, then your background is going to be “more in focus” and the field curvature will be fighting against you.
1. Turbofrog · March 8, 2018
  Yes, I’m curious to have this clarified as well.
2. Brandon Dube · March 8, 2018
  A positive number on the field curvature plot indicates the best focus is closer to the lens. Closer to the lens is focused in the => infinity direction, or beyond infinity in this case because the “0” is already at infinity.
  1. Samuel H · March 8, 2018
    OK, so in this case field curvature does work for you in portraits. Thanks!
    1. Brandon Dube · March 8, 2018
      To be sure you interpreted what I wrote as intended — this means that if you focus on a face in the center, the background at the edge is more in focus. That is what “=> infinity direction” is meant to convey.
      1. Samuel H · March 9, 2018
        Argh, then NO, field curvature is working AGAINST you!!!
        
        Thanks again <3
      2. Clayton Taylor · March 9, 2018
        So, shooting with a wide-open aperture, the family in the center of the frame posing with Mickey Mouse is at the point of focus, and the Magic Kingdom castle and other fantasy structures at the edges will be in better focus than if the lens design had a flat-field. Genius.
    2. Brandon Dube · March 8, 2018
      To be sure you interpreted what I wrote as intended -- this means that if you focus on a face in the center, the background at the edge is more in focus. That is what "=> infinity direction" is meant to convey.
      1. Samuel H · March 9, 2018
        Argh, then NO, field curvature is working AGAINST you!!!
        Thanks again <3
3. Jared M Johnson · March 8, 2018
  Field curvature rarely ever curves away from the lens, right?
  I mean the field usually curves towards a field equidistant from the aperture, like a sphere around the camera.
  It would be weird for it to focus in a field curving away…
  1. Ilya Zakharevich · March 9, 2018
    Exactly the opposite. If your subject is a sphere of radius 10cm (like a face), then with suitable field curvature, all the visible surface will be in focus. (No matter what is the subject distance. 😉
    1. Roger Cicala · March 9, 2018
      This gave me one of those ‘of course’ moments. Now I’m going to get one of those lenses and see if I can actually accomplish this thing!!! 🙂
      1. Ilya Zakharevich · March 10, 2018
        Keep in mind that for the diagrams above, radius of curvature in the object space is about 20cm for tangential, and about 40cm for saggital. So to get 10cm, you need something with more dramatic curvature!
        
        ????????????????????????????????????????
        
        How to to see this starting from your 2d colormaps: rescale them 20× w.r.t. HORIZONTAL markings (so that the width becomes 36cm instead of 18mm). Place them horizontally so that the vertical midline goes along the optical axis, and so that the horizontal midline is “the subject”. Now it is a color map of MTF in the target space provided the field of view is 36cm wide.
        
        (The only ”magic number” in the recipe above is the factor 20. This is the ratio of scales on the vertical and horizontal axes of your colormaps.)
        
        Brandon Dube · March 11, 2018
        The field curvature could potentially be very different at a closer distance to at infinity. Especially in fast internal focus lenses.
    2. Jared M Johnson · March 9, 2018
      That’s definitely not right.
      That’s not how uncorrected lenses work.
      If you took a picture of the inside of a hemisphere it would be in focus, not the outside.
      1. Jan Stenberg · March 9, 2018
        I understand how you think even though I believe it would be hard to get focus of the outside of the sphere from the inside of a sphere. 😉 Maybe it would be true for an uncorrected lens, but these system of lenses are probably corrected for many types of abberations. My personal experience of a lens with that “feature” is the Voigtlander 21mm f/1.8 which on a full frame camera would do just that, render pictures with both nose and ears sharp at the right distance to the subject. 🙂
        
        Valentin Alexiss · March 11, 2018
        That’s it : the Oly 45mm 1.8 does this too. It give’s focus for the sphere from the outside of it. So a noze focus will give more focus to ears… than with a pur « planar » design.
        
        mrclean808 · January 6, 2019
        jan_stenberg briefly
Samuel H · March 8, 2018
I have a doubt about your statement that "for portraits, I’d probably want the Zuiko Digital f/1.8; that field curvature will help throw everything other than the subject out of focus".
I hope this means that the curve in these graphs means the sides are focused closer than the center is. If it means that the center is focusing closer than the sides, then your background is going to be "more in focus" and the field curvature will be fighting against you.
1. Turbofrog · March 8, 2018
  Yes, I'm curious to have this clarified as well.
2. Brandon Dube · March 8, 2018
  A positive number on the field curvature plot indicates the best focus is closer to the lens. Closer to the lens is focused in the => infinity direction, or beyond infinity in this case because the "0" is already at infinity.
3. Jared M Johnson · March 8, 2018
  Field curvature rarely ever curves away from the lens, right?
  I mean the field usually curves towards a field equidistant from the aperture, like a sphere around the camera.
  It would be weird for it to focus in a field curving away...
  1. Ilya Zakharevich · March 9, 2018
    Exactly the opposite. If your subject is a sphere of radius 10cm (like a face), then with suitable field curvature, all the visible surface will be in focus. (No matter what is the subject distance. ;-)
    1. Roger Cicala · March 9, 2018
      This gave me one of those 'of course' moments. Now I'm going to get one of those lenses and see if I can actually accomplish this thing!!! :-)
      1. Ilya Zakharevich · March 10, 2018
        Keep in mind that for the diagrams above, radius of curvature in the object space is about 20cm for tangential, and about 40cm for saggital. So to get 10cm, you need something with more dramatic curvature!
        ────────────────────────────────────────
        How to to see this starting from your 2d colormaps: rescale them 20× w.r.t. HORIZONTAL markings (so that the width becomes 36cm instead of 18mm). Place them horizontally so that the vertical midline goes along the optical axis, and so that the horizontal midline is “the subject”. Now it is a color map of MTF in the target space provided the field of view is 36cm wide.
        (The only ”magic number” in the recipe above is the factor 20. This is the ratio of scales on the vertical and horizontal axes of your colormaps.)
    2. Jared M Johnson · March 9, 2018
      That's definitely not right.
      That's not how uncorrected lenses work.
      If you took a picture of the inside of a hemisphere it would be in focus, not the outside.
      1. Jan Stenberg · March 9, 2018
        I understand how you think even though I believe it would be hard to get focus of the outside of the sphere from the inside of a sphere. ;-) Maybe it would be true for an uncorrected lens, but these system of lenses are probably corrected for many types of abberations. My personal experience of a lens with that "feature" is the Voigtlander 21mm f/1.8 which on a full frame camera would do just that, render pictures with both nose and ears sharp at the right distance to the subject. :-) The focus is rapped around the head (concave?) rather than flat or convex. Don't ask me How it works, I just know that's a "great" portrait lens. ;-)
        Valentin Alexiss · March 11, 2018
        That's it : the Oly 45mm 1.8 does this too. It give's focus for the sphere from the outside of it. So a noze focus will give more focus to ears... than with a pur « planar » design.
cb5107 · March 8, 2018
Thanks for a great article. This helps me understand why I chose the P42.5/1.7 over the O45/1.8 based on my testing using architectural (i.e. stone or brick walls) type shots.
In general, it looks like you cannot lose with any of these lenses.
Hope you find time to look at the 17mm lenses too :).
Ketan Gajria · March 8, 2018
Is it possible for different copies of the same lens to have different curvatures of field? I bought the best of 4 copies tested for the Olympus 25mm f/1.2 but the outer image regions don’t really improve from f/1.2 to ~f/4 while the center does and it boggles my mind. My Sony Zeiss 55mm f/1.8 FE is similar.
1. Roger Cicala · March 8, 2018
  Ketan, it is definitely possible. Look back at the graphs above and you’ll see some mildly tilted fields. Some are more tilted, some the field is decentered.
Ketan Gajria · March 8, 2018
Is it possible for different copies of the same lens to have different curvatures of field? I bought the best of 4 copies tested for the Olympus 25mm f/1.2 but the outer image regions don't really improve from f/1.2 to ~f/4 while the center does and it boggles my mind. My Sony Zeiss 55mm f/1.8 FE is similar.
1. Roger Cicala · March 8, 2018
  Ketan, it is definitely possible. Look back at the graphs above and you'll see some mildly tilted fields. Some are more tilted, some the field is decentered.
Phanter · March 8, 2018
Just a theoretical question. If you test FF lenses at 10,20,30,40,50 lp/mm shouldn’t you test APS-C lenses at 15,30,45,60,75 lp/mm and M43 lenses at 20,40,60,80,100 lp/mm? Different frequencies describe different things on different sensor formats. So why don’t you test with constant lp/ph?
1. Samuel H · March 8, 2018
  I guess it’s easier to use the same settings in all tests, but you’re right that you have to compare the 40 lp/mm here with 20 lp/mm in a FF lens. Which I did, because I wanted to see if any of these could be an alternative to, say, a Sony FE 85m f/1.8. My conclusion: no small sensors for me, thanks.
2. Roger Cicala · March 8, 2018
  Technically yes. But I’d have to put a 5,000 word explanation along with it; and m4/3 fanboys would have a meltdown over it and well, just not gonna do it 🙂
  1. Phanter · March 8, 2018
    I understand, but it would be nice, if you at least tested those higher frequencies. Also i think if you can find the time you should write those 5000 words, this way you could put an end to all this discussion.
Phanter · March 8, 2018
Just a theoretical question. If you test FF lenses at 10,20,30,40,50 lp/mm shouldn't you test APS-C lenses at 15,30,45,60,75 lp/mm and M43 lenses at 20,40,60,80,100 lp/mm? Different frequencies describe different things on different sensor formats. So why don't you test with constant lp/ph?
1. Samuel H · March 8, 2018
  I guess it's easier to use the same settings in all tests, but you're right that you have to compare the 40 lp/mm here with 20 lp/mm in a FF lens. Which I did, because I wanted to see if any of these could be an alternative to, say, a Sony FE 85m f/1.8. My conclusion: no small sensors for me, thanks.
2. Roger Cicala · March 8, 2018
  Technically yes. But I'd have to put a 5,000 word explanation along with it; and m4/3 fanboys would have a meltdown over it and well, just not gonna do it :-)
  1. Phanter · March 8, 2018
    I understand, but it would be nice, if you at least tested those higher frequencies. Also i think if you can find the time you should write those 5000 words, this way you could put an end to all this discussion.
Andreas Jordan · March 8, 2018
Tne Panasonic Leica lens is called Noctricron not Noctritron.
1. Roger Cicala · March 8, 2018
  Hey they make the name that damn long, they gonna get some people not paying much attention to how you spell it. 🙂
  1. Hank Roest · March 9, 2018
    So right, Roger. It’s actually not Noctricron either, Andreas. You got an extra “r” in there, buddy. ;-))
  2. Athanasius Kirchner · March 9, 2018
    Noctritron… wasn’t that a brand of TV’s made by Sony? Maybe they’re the ones behind this lens!? /jk
Andreas Jordan · March 8, 2018
Tne Panasonic Leica lens is called Noctricron not Noctritron.
1. Roger Cicala · March 8, 2018
  Hey they make the name that damn long, they gonna get some people not paying much attention to how you spell it. :-)
  1. Hank Roest · March 9, 2018
    So right, Roger. It's actually not Noctricron either, Andreas. You got an extra "r" in there, buddy. ;-))
  2. Athanasius Kirchner · March 9, 2018
    Noctritron... wasn't that a brand of TV's made by Sony? Maybe they're the ones behind this lens!? /jk
Stefanie Daniella · March 8, 2018
lol
i thought the
XiaoYi 42.5mm f1.8
m43 lens (the kit prime for Yi M1 to avoid)
would break the “worst lens = longest name” rule
(it has terrible CA according to some)
but it wasn’t tested
the kit 12-40 zoom for Yi M1 is more tolerable
Stefanie Daniella · March 8, 2018
lol
i thought the
XiaoYi 42.5mm f1.8
m43 lens (the kit prime for Yi M1 to avoid)
would break the "worst lens = longest name" rule
(it has terrible CA according to some)
but it wasn't tested
the kit 12-40 zoom for Yi M1 is more tolerable
s.wolters · March 9, 2018
Thank you for this solid review. I will dig in your archives to consume some more. A few remarks though.
– The Olympus 45mm f/1.8 is not even close to $400. In fact you will find it everywhere in the world far below $300.
– The most obvious and visible differences between these lenses can be noticed when comparing the final images. Panasonic usually aims at a higher contrast and the colors are different from Olympus. Personally I find the images too hard and the colors are a bit out of balance. This keeps me in the Olympus camp, certainly when buying cameras.
(Minor error: the Olympus PRO is a 45mm, not 42,5mm)
1. Roger Cicala · March 9, 2018
  Thank you. I just grab the B&H prices, so it’s good to know it’s worth comparison shopping these.
s.wolters · March 9, 2018
Thank you for this solid review. I will dig in your archives to consume some more. A few remarks though.
- The Olympus 45mm f/1.8 is not even close to $400. In fact you will find it everywhere in the world far below $300.
- The most obvious and visible differences between these lenses can be noticed when comparing the final images. Panasonic usually aims at a higher contrast and the colors are different from Olympus. Personally I find the images too hard and the colors are a bit out of balance. This keeps me in the Olympus camp, certainly when buying cameras.
(Minor error: the Olympus PRO is a 45mm, not 42,5mm)
1. Roger Cicala · March 9, 2018
  Thank you. I just grab the B&H prices, so it's good to know it's worth comparison shopping these.
Goblin · March 9, 2018
Blast !!! The mystery of the yellow paint tube – finally solved ???

Long live Mr Cicala 🙂

So. About that yellow paint tube. Let me tell you a tale

One upon of time, in a kingdom far far away (kingdom moved from UK to Seattle at some point) called DPReview, there was this beautiful, helpful, easy to use Studio Scene.

On said scene, some bored pixel peeper (me), noticed something disturbing…

ALL the m43 cameras reviewed after the Olympus E-M5 mk1 (that was in 2012) looked worse than said E-M5 mk1.
Especially flagrant was the yellow paint tube on the lower right. Ah, that yellow paint tube !!! It looked fuzzier and worse on anything after that E-M5 mk1. Panny bodies, Oly bodies, you name it.

Then, somewhere along the tests, DPR’s reviewers kindly shared that the 45mm f:1.8 M.Zuiko, although “not as good” as the 50mm f:2 4/3 Zuiko Macro previously used, was “good enough” to become the new reference lens after 2012. And all further tests were shot with it.

And that thing has – wait – field curvature ? And the paint tube is on the corner of the studio scene… And the center of the scene is fine…

The difference was once more highlighted during the test of the then new Pixel Shift feature on the then new E-M5 mk2. For the Pixel Shift tests only, the Panaleica 42.5mm f:1.2 was used, because they “felt” a better lens than the 45mm f:1.8 was needed for that one. And the difference was there.

Funnily enough, when testing the Pixel Shift option on the E-M1 mk2 later, the 45mm f:1.8 lens was used even for the Pixel Shift pictures.

Want to bet what the yellow paint tube looks like on the Pixel Shift images between the E-M5 mk2 and E-M1 mk2, on that studio scene ?

Check it out 🙂

I have them all except the Zuiko 45mm f:1.2. Pretty please, can someone give me an excuse to get it as well ?

Cheers
1. Roger Cicala · March 9, 2018
  That was awesome. And now I have to post this PS thing Joey did of Aaron and me a while back. https://uploads.disquscdn.com/images/3df3e908d47bab342c8a48c16697f509811aae9d16577503c45c1ba6cb422c95.jpg
  1. Goblin · March 13, 2018
    Blast, I’ve seen too many “…Lenin working with scientists for a glowing future…” posters in my youth for this not to shiver me, timbers or not 😀
2. Bryce Steiner · August 19, 2018
  I noticed this exact issue and let DP know, but they won’t do anything about the bad lens being used.
Goblin · March 9, 2018
Blast !!! The mystery of the yellow paint tube - finally solved ???
Long live Mr Cicala :)
So. About that yellow paint tube. Let me tell you a tale
One upon of time, in a kingdom far far away (kingdom moved from UK to Seattle at some point) called DPReview, there was this beautiful, helpful, easy to use Studio Scene.
On said scene, some bored pixel peeper (me), noticed something disturbing...
ALL the m43 cameras reviewed after the Olympus E-M5 mk1 (that was in 2012) looked worse than said E-M5 mk1.
Especially flagrant was the yellow paint tube on the lower right. Ah, that yellow paint tube !!! It looked fuzzier and worse on anything after that E-M5 mk1. Panny bodies, Oly bodies, you name it.
Then, somewhere along the tests, DPR's reviewers kindly shared that the 45mm f:1.8 M.Zuiko, although "not as good" as the 50mm f:2 4/3 Zuiko Macro previously used, was "good enough" to become the new reference lens after 2012. And all further tests were shot with it.
And that thing has - wait - field curvature ? And the paint tube is on the corner of the studio scene... And the center of the scene is fine...
The difference was once more highlighted during the test of the then new Pixel Shift feature on the then new E-M5 mk2. For the Pixel Shift tests only, the Panaleica 42.5mm f:1.2 was used, because they "felt" a better lens than the 45mm f:1.8 was needed for that one. And the difference was there.
Funnily enough, when testing the Pixel Shift option on the E-M1 mk2 later, the 45mm f:1.8 lens was used even for the Pixel Shift pictures.
Want to bet what the yellow paint tube looks like on the Pixel Shift images between the E-M5 mk2 and E-M1 mk2, on that studio scene ?
Check it out :)
I have them all except the Zuiko 45mm f:1.2. Pretty please, can someone give me an excuse to get it as well ?
Cheers
1. Roger Cicala · March 9, 2018
  That was awesome. And now I have to post this PS thing Joey did of Aaron and me a while back. https://uploads.disquscdn.c...
  1. Goblin · March 13, 2018
    Blast, I've seen too many "...Lenin working with scientists for a glowing future..." posters in my youth for this not to shiver me, timbers or not :-D
2. Bryce Steiner · August 19, 2018
  I noticed this exact issue and let DP know, but they won't do anything about the bad lens being used.
3. spider-mario · October 8, 2019
  I have them all except the Zuiko 45mm f:1.2. Pretty please, can someone give me an excuse to get it as well ?
  Gorgeous “feathered” bokeh?
  https://photos.smugmug.com/...
  https://photos.smugmug.com/...
Horst Kath · March 10, 2018
Hello Roger, It was very interesting to read your publication about 4/3 mft lenses. The most surprising thing for me is the value of deviation between the same lens but different copies. As I read your review about the 45 mm lenses I found that there must be a mistake in comparing the zuiko 45 / 1.8 at an aperture of 1.8 and 2.8. I can’t believe that the performance at an aperture of 2.8 is so much worth than at 1.8. May be the charts has been exchanged? Bests regards Horst
Bonvil · March 10, 2018
Hello Roger, It was very interesting to read your publication about 4/3 mft lenses. The most surprising thing for me is the value of deviation between the same lens but different copies. As I read your review about the 45 mm lenses I found that there must be a mistake in comparing the zuiko 45 / 1.8 at an aperture of 1.8 and 2.8. I can't believe that the performance at an aperture of 2.8 is so much worth than at 1.8. May be the charts has been exchanged? Bests regards Horst
Hendrik · March 10, 2018
Re: Handling copy-to-copy variation

Thanks Roger for yet another great article! The copy-to-copy variation in m4/3 lenses you’ve mentioned before and it is a little worrying. I wonder which procedure you would recommend (or maybe there is a post here about it already) for the home user to choose among serveral specimen of a lens?
1. Roger Cicala · March 11, 2018
  Hendrick, I recommend a day taking pictures and then spending a little time looking at 50% magnification images. If something is really wrong with the lens, you’ll notice it.
Hendrik · March 10, 2018
Re: Handling copy-to-copy variation
Thanks Roger for yet another great article! The copy-to-copy variation in m4/3 lenses you've mentioned before and it is a little worrying. I wonder which procedure you would recommend (or maybe there is a post here about it already) for the home user to choose among serveral specimen of a lens?
1. Roger Cicala · March 11, 2018
  Hendrick, I recommend a day taking pictures and then spending a little time looking at 50% magnification images. If something is really wrong with the lens, you'll notice it.
Catherine Mann · March 10, 2018
So which is best as legacy lens? I would think Voigtlander. Is their customer service for repair not the same as pre digital lenses, thinking Mamiya medium format. Went ahead and got the Voigtlander as saving for Lumix G9 during slow winter season. It is so clear but first time heard about repairs. Like idea having lens all manual focus so don’t feel lose investment when af tech changes. Also discovered Austrian heritage and liked company was first sponsored there. Take time to craft for quality.
Catherine Mann · March 10, 2018
So which is best as legacy lens? I would think Voigtlander. Is their customer service for repair not the same as pre digital lenses, thinking Mamiya medium format. Went ahead and got the Voigtlander as saving for Lumix G9 during slow winter season. It is so clear but first time heard about repairs. Like idea having lens all manual focus so don't feel lose investment when af tech changes. Also discovered Austrian heritage and liked company was first sponsored there. Take time to craft for quality.
Amin · March 11, 2018
“Here’s a great example of why we have internet arguments. Two of these eight lenses are better than any of the Olympus Pros; three are worse. Reviewers or photographers who compared one copy of each are probably going to draw different conclusions.”

Such an important point to keep in mind. Copy variation is and always has been a fact of life with all manufacturers and formats. It’s also notable that even the good copies which have a center of sharpness in the center of the lens tend to have some degree of asymmetric sharpness which is noticeable if you go looking for it. I try NOT to go looking for it :).
Amin · March 11, 2018
"Here’s a great example of why we have internet arguments. Two of these eight lenses are better than any of the Olympus Pros; three are worse. Reviewers or photographers who compared one copy of each are probably going to draw different conclusions."
Such an important point to keep in mind. Copy variation is and always has been a fact of life with all manufacturers and formats. It's also notable that even the good copies which have a center of sharpness in the center of the lens tend to have some degree of asymmetric sharpness which is noticeable if you go looking for it. I try NOT to go looking for it :).
tom · March 11, 2018
… Once again my question …

Hello… Great work again and actually the best technical evaluation of optics around the Web…

In his publication i wonder whether the curves of the Olympus 45 1.2 and Pana 42.5 1.2 for ‘ 1.2 ‘ are different in the charts at the beginning and when stopped down… my personal experience when using both: the latter curves seem to be more realistic because the olympus really is sharper already in the center and especially in the corners

Which graphics are correct?
1. David Bateman · March 11, 2018
  This is an easy question the answer is they are both correct. The top mtf curves are an average of 6 to 10 lenses. The stop down curves are a single lens. The single Panasonic stop down curve is a less than average performer. The single Olympus stop down curve is an average performer.
  What is concerning is the sample viariation in M43rds lenses.
  1. tom · March 11, 2018
    Hello David and thx a lot for your interesting and quick answer;-)!
    .. what you are describing -unfortunately- remembers me of my 25 1.2 experience: from 4 examples only one was top, two were average and one was really bad over the complete field from center to edges… and what was even much more disappointing: also after “servicing” this lens by Oly. the optical quality was still rather bad… so i only kept the “best” .. and when i bought the 17 1.2 i also had the chance to compare 2 examples; here the variations were very small…(but the focus ring on both was a bit loose and shows some minor slippage (other than the 25 and 45..). But at the end i will keep all these great lenses because of their really fantastic image overall image “effects”: a comprehensive approach to integrate ALL pictorial effects for an aesthetical image … so the sheer will or intention of Olympus in todays competitive markets has to be appreciated (for example in comparison the the Sony approach with “normal” lenses, G-Master optics and Zeiss glass….) cheers, Tom
  2. Brandon Dube · March 11, 2018
    We select an “average” lens for the stop down.
    1. David Bateman · March 11, 2018
      Brandon, I say the selected Panasonic lens is below average, because it clearly is. Look at the average plot of 10 lenses, the 50lp/mm center value is at 0.5, whereas the selected lens is at 0.4. However, based on the variation plots, I understand that the selected Panasonic maybe at the mode.
      To understand the difference, imagine a class of 10 students takes a test. 2 get 100%, 1 gets 45%, and 7 get 65%. The average is 70%, so 8 out of 10 kids are below average, whereas the mode for the class is 65.
      1. Brandon Dube · March 12, 2018
        I do not understand why people devote such incredible attention to the on-axis point. By area, it is the least significant. It is easy to visually reference because it lies on an axis bound, but it is not as special as, say, looking at the center 30% by area.
        
        The copy selected for stop-down is a bit below average on-axis, but is closer to the median because a copy or two is extremely good while the “average” is less special. The variance in this model is actually quite extreme.
        
        This copy was selected for having the least perturbed astigmatism as a function of field, despite having a bit worse performance on-axis. The astigmatism being perturbed will substantially (negatively) impact the closed aperture performance, which is unrealistic. We look for more than just matching MTF numbers when picking the stop down copy.
        
        David Bateman · March 12, 2018
        Brandon, I hope I didn’t come off as being rude. If I did I am sorry as that was never my intent. What you are saying is what I was saying. As I tried to give an example. Since I don’t have the data, I don’t know the median value. However, based on the variation I could infer the mode, the most common same which I do think you presented.
tom · March 11, 2018
... Once again my question ...
Hello... Great work again and actually the best technical evaluation of optics around the Web...
In his publication i wonder whether the curves of the Olympus 45 1.2 and Pana 42.5 1.2 for ' 1.2 ' are different in the charts at the beginning and when stopped down... my personal experience when using both: the latter curves seem to be more realistic because the olympus really is sharper already in the center and especially in the corners
Which graphics are correct?
1. David Bateman · March 11, 2018
  This is an easy question the answer is they are both correct. The top mtf curves are an average of 6 to 10 lenses. The stop down curves are a single lens. The single Panasonic stop down curve is a less than average performer. The single Olympus stop down curve is an average performer.
  What is concerning is the sample viariation in M43rds lenses.
  1. tom · March 11, 2018
    Hello David and thx a lot for your interesting and quick answer;-)!
    .. what you are describing -unfortunately- remembers me of my 25 1.2 experience: from 4 examples only one was top, two were average and one was really bad over the complete field from center to edges... and what was even much more disappointing: also after "servicing" this lens by Oly. the optical quality was still rather bad... so i only kept the "best" .. and when i bought the 17 1.2 i also had the chance to compare 2 examples; here the variations were very small...(but the focus ring on both was a bit loose and shows some minor slippage (other than the 25 and 45..). But at the end i will keep all these great lenses because of their really fantastic image overall image "effects": a comprehensive approach to integrate ALL pictorial effects for an aesthetical image ... so the sheer will or intention of Olympus in todays competitive markets has to be appreciated (for example in comparison the the Sony approach with "normal" lenses, G-Master optics and Zeiss glass....) cheers, Tom
  2. Brandon Dube · March 11, 2018
    We select an "average" lens for the stop down.
    1. David Bateman · March 11, 2018
      Brandon, I say the selected Panasonic lens is below average, because it clearly is. Look at the average plot of 10 lenses, the 50lp/mm center value is at 0.5, whereas the selected lens is at 0.4. However, based on the variation plots, I understand that the selected Panasonic maybe at the mode.
      To understand the difference, imagine a class of 10 students takes a test. 2 get 100%, 1 gets 45%, and 7 get 65%. The average is 70%, so 8 out of 10 kids are below average, whereas the mode for the class is 65.
      1. Brandon Dube · March 12, 2018
        I do not understand why people devote such incredible attention to the on-axis point. By area, it is the least significant. It is easy to visually reference because it lies on an axis bound, but it is not as special as, say, looking at the center 30% by area.
        The copy selected for stop-down is a bit below average on-axis, but is closer to the median because a copy or two is extremely good while the "average" is less special. The variance in this model is actually quite extreme.
        This copy was selected for having the least perturbed astigmatism as a function of field, despite having a bit worse performance on-axis. The astigmatism being perturbed will substantially (negatively) impact the closed aperture performance, which is unrealistic. We look for more than just matching MTF numbers when picking the stop down copy.
        David Bateman · March 12, 2018
        Brandon, I hope I didn't come off as being rude. If I did I am sorry as that was never my intent. What you are saying is what I was saying. As I tried to give an example. Since I don't have the data, I don't know the median value. However, based on the variation I could infer the mode, the most common same which I do think you presented.
David Bateman · March 11, 2018
Question, is it hard to design lenses with the field curvature going in the opposite direction as what you have presented for the Olympus f1.8 lens?
I was thinking about many photographs and rarely is there lots of stuff in front of the model. To quiet the razer thin 135 format people, a lens designed that way would give you an apparent thin DOF, as long as your object of focus is not blossoms on a tree.
1. Brandon Dube · March 11, 2018
  Yes, it is very difficult to do that. Consider an object at infinity, no matter the angle of view the object is the same distance away. The image, then, wants to be an equal distance from the lens, along a sphere with radius of the focal length, not a plane. To make the field curvature go the other way, you need the sum of the power of the negative elements in your lens to be bigger than the sum of the power of the positive elements in your lens.
  
  Doing that without having a negative focal length is hard. Doing it with a positive focal length and the rest of the aberrations well-corrected is harder.
  1. David Bateman · March 11, 2018
    Thank you Brandon,
    That makes sense. Sounds similar to waiting for a metalens with a negative refractive index to transmit the evanescent spectrum. Not likely to occur in my lifetime.
David Bateman · March 11, 2018
Question, is it hard to design lenses with the field curvature going in the opposite direction as what you have presented for the Olympus f1.8 lens?
I was thinking about many photographs and rarely is there lots of stuff in front of the model. To quiet the razer thin 135 format people, a lens designed that way would give you an apparent thin DOF, as long as your object of focus is not blossoms on a tree.
1. Brandon Dube · March 11, 2018
  Yes, it is very difficult to do that. Consider an object at infinity, no matter the angle of view the object is the same distance away. The image, then, wants to be an equal distance from the lens, along a sphere with radius of the focal length, not a plane. To make the field curvature go the other way, you need the sum of the power of the negative elements in your lens to be bigger than the sum of the power of the positive elements in your lens.
  Doing that without having a negative focal length is hard. Doing it with a positive focal length and the rest of the aberrations well-corrected is harder.
  1. David Bateman · March 11, 2018
    Thank you Brandon,
    That makes sense. Sounds similar to waiting for a metalens with a negative refractive index to transmit the evanescent spectrum. Not likely to occur in my lifetime.
Justin · March 14, 2018
“and if you think a metal housing makes lenses more reliable, come spend a day in our repair shop” This was interesting to me both for the funny tone and for how it goes against “conventional wisdom” – do you find that housing material doesn’t matter for durability? I ask because I have a set of Rokinon Cine DS lenses and the plastic housing always worries me, despite them working reliably for several years now.
1. Roger Cicala · March 15, 2018
  It definitely doesn’t make them more reliable, and in some ways they are more frequent repairs. There’s no statistically significant difference in repair rate, but metal is, if anything, slightly higher.
  1. Justin · March 15, 2018
    Huh! Interesting!
  2. Justin · August 30, 2018
    Huh – fascinating. Thank you. 🙂
NoGodsNoBosses · March 14, 2018
"and if you think a metal housing makes lenses more reliable, come spend a day in our repair shop" This was interesting to me both for the funny tone and for how it goes against "conventional wisdom" - do you find that housing material doesn't matter for durability? I ask because I have a set of Rokinon Cine DS lenses and the plastic housing always worries me, despite them working reliably for several years now.
1. Roger Cicala · March 15, 2018
  It definitely doesn't make them more reliable, and in some ways they are more frequent repairs. There's no statistically significant difference in repair rate, but metal is, if anything, slightly higher.
  1. NoGodsNoBosses · March 15, 2018
    Huh! Interesting!
  2. NoGodsNoBosses · August 30, 2018
    Huh - fascinating. Thank you. :)
Najinsky · May 18, 2018
Thanks for yet another excellent article.

Did I detect higher frequency humor and a softening of the jibes?

You know you’re supposed to be getting more cantankerous right? A kind of field curvature as we advance towards the edges.
Najinsky · May 18, 2018
Thanks for yet another excellent article.
Did I detect higher frequency humor and a softening of the jibes?
You know you’re supposed to be getting more cantankerous right? A kind of field curvature as we advance towards the edges.
Edit: forgot to say, love the new format diagrams for field flatness.
DrJon · May 18, 2018
An interesting question about the expensive Panasonics is whether they can repair them locally or will swap it for a refurb as they have to go to Japan to get sorted. This appears to be potentially excitingly expensive if out of warranty. This post/thread sums up one person’s problem (with the 100-400, CDN$1700 for a swap-out):
https://www.dpreview.com/forums/post/61158773
DrJon · May 18, 2018
An interesting question about the expensive Panasonics is whether they can repair them locally or will swap it for a refurb as they have to go to Japan to get sorted. This appears to be potentially excitingly expensive if out of warranty. This post/thread sums up one person's problem (with the 100-400, CDN$1700 for a swap-out):
https://www.dpreview.com/fo...
Greg Rook · May 29, 2018
Thanks very much for this wonderful and detailed report. My personal experience with M43 lenses, if you are a nit-picking perfectionist like I am, both a master optician and photographer, is that if you simply can’t live without the best quality, if doubt bugs you and keeps you up at night, then you absolutely have to test every lens you buy. BTW, I am aware that M43 is mostly targeted to the advanced and prosumer market, and they believe that most amateurs really won’t notice, or care. Really, most all lenses take really “good” pictures, even if they are the lenses on your smart phone. But for people like me, I am sure manufacturers don’t someone spreading the word, because then everybody would be returning 3 out of 4 lenses they buy! There really can be big variations between samples, not just at the long edges and corners! Sometimes the lenses are softer overall, or they can be just concentrically soft radiating outwards. Remarkably, some are even soft in just in the center! Thank goodness that most retailers are understanding and will accept exchanges!
Brian Caldwell · July 7, 2018
But….you forgot the sharpest (and admittedly the biggest and most expensive) “40” of all: Zeiss 55/1.4 + Speed Booster Ultra = 39mm f/1.0. IIRC, this had measured ~80% contrast at 30 cycles @ f/1 when the correct thickness of glass was placed in the measurement path.
Brian Caldwell · July 7, 2018
But....you forgot the sharpest (and admittedly the biggest and most expensive) "40" of all: Zeiss 55/1.4 + Speed Booster Ultra = 39mm f/1.0. IIRC, this had measured ~80% contrast at 30 cycles @ f/1 when the correct thickness of glass was placed in the measurement path.
Ruairi Winter Johnston · February 12, 2020
Does temperature account for some copy variance? I’m thinking of my solid neck guitar, if I tune it at midday, it’ll be out of tune when the temperature drops at night, but go back in tune by midday again.
1. Roger Cicala · February 12, 2020
  It could, but all our testing is done at the same temperature.
Ergellegre · February 12, 2020
Does temperature account for some copy variance? I'm thinking of my solid neck guitar, if I tune it at midday, it'll be out of tune when the temperature drops at night, but go back in tune by midday again.

Finally, Some More m4/3 MTF Testing: Are the 40s Fabulous?

Let’s Meet Today’s Players

MTF TESTS

Panasonic Leica DG Macro Elmarit 45mm f/2.8 (8 copies only)

Panasonic Lumix 42.5mm f1.7 ASPH OIS (6 copies)

Stop-Down Tests

Voigtlander at f/1.4

All Lenses at f/2.8

Field Curvature (MTF vs. Field vs. Focus)

Copy-to-Copy Variation

So What Did We Learn Today?

Addendum: Field Curvature and MTF

Tags

223 Comments