Are m43 lenses really so good?

Discussion in 'Native Lenses' started by Klorenzo, Jun 28, 2015.

  1. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    Out of curiosity I went to the slrgear site to check the 50/1.8 lens performance. Yes, it a cheap basic lens but it is often considered a good lens. Well, I've never seen a chart so ugly for any m43 lens!

    http://slrgear.com/reviews/showproduct.php/product/150/cat/10

    You have to stop it down to f4 or f5.6 to get performances similar to the Pana and Oly kit zoom:

    http://slrgear.com/reviews/showproduct.php/product/1400/cat/15
    http://slrgear.com/reviews/showproduct.php/product/1588/cat/69

    So I checked the Canon 24-70 and is not bad but it is debatable if it is any better or worse then the Pana and Oly pro zooms.
    Then the Canon 85/1.2 and the chart is quite horrible, not even comparable to the Nocticron. Nikon 105/2.8 is terrible. I'm talking about what I see in the "blur index" chart, I have not interest in bashing other systems for the sake of it.
    So I went to the Otus 55(!) and again it never reaches the sharpness of the Oly 75 and f4 ever.

    So I'm wondering: is this true or is just some measurement "effect" (they use DxO software BTW). Is the different format ratio that is helping? The smaller surface to cover? Shorter flange distance? Less telecentric designs? To make it simple: are m43 lenses really so much better?

    Ok, lens performance is not only about sharpness, there is CA, distortions, etc. but let's talk only about the slrgear "blur index" for now.
     
    Last edited: Jun 28, 2015
  2. cyrax83

    cyrax83 Mu-43 Regular

    193
    Dec 8, 2014
    Smaller area of light to resolve so perhaps easier to make better optics?
     
  3. pdk42

    pdk42 One of the "Eh?" team

    Jan 11, 2013
    Leamington Spa, UK
    M43 lenses need to be sharp since the pixel density in m43 sensors is high and of course the image needs greater magnification than a 135 frame. Having said that, you're comparing m43 lenses against old Canikon designs. Do it with their newer lenses and you'll get a different story.
     
  4. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    I think the 24-70 2.8 II is a recent lens, isn't it?

    http://slrgear.com/reviews/showproduct.php/product/1494/cat/11

    And what about the Otus? Anyway can you point me to a couple of recent Canikon lenses that I can take a loot at?

    Regarding pixel density: do you mean that m43 are actually able to optically resolve more or that the lower DSLR pixel density of the sensor limits what the test is able to verify?
     
  5. StefanKruse

    StefanKruse Mu-43 Veteran

    349
    Jan 28, 2015
    Denmark
    Stefan
    Interesting analysis - Im afraid I cant add anything to the discussion as I have only been shooting for a year, add to my lack of expereince the fact that I have no clue of how to use/read these strange charts :)

    Anyway I appreciate the analysis and discussion - Keep it going.
     
  6. DaveEP

    DaveEP Mu-43 Top Veteran

    684
    Sep 20, 2014
    York, UK
    One of the major advantages of M43 is that the lenses 'can' be quite good wide open. Of the M43 lenses I've tested only the Olympus 45mm has really disappointed me wide open, though it can also depend on the distance to the subject (Olympus 45mm Review). The closer the subject the better it seems to be. Distant subjects (using my normal comparison shots) were less than impressive.

    Most other M43 lenses from Panasonic and Olympus do appear to be quite good wide open, though as has been discussed elsewhere on this forum, the Voightlander f0.95 aren't always so impressive.

    One of the major frustrations I had with FF 35mm lenses was how soft they were wide open. Many of the Canon / Nikon lenses were horrible, and yes I've had a lot of f1.2 and f1.4 primes and sold them all. While they were OK in studio environments, get them out in to the wild and you have to be really careful how you use them.
     
  7. pdk42

    pdk42 One of the "Eh?" team

    Jan 11, 2013
    Leamington Spa, UK
    Ah, when you said 24-70, I assumed you meant the original, not the mk II. That is indeed a recent lens and considered to be amongst the best. I'd be surprised if it's beaten by any current m43 lens.

    As regards pixel density, the theoretical resolution of the sensor increases as the pixel size reduces. Clearly, if you keep reducing pixel size there will be a point at which a lens will resolve less than the sensor (or in extremist we hit the diffraction limit). To make best use of a sensor with higher pixel density, you need a lens capable of higher resolution. Of course, if the overall sensor size is smaller, then you need to enlarge the image more to gain the same end result as a bigger sensor. The net result is that to get equivalently sharp images on a smaller, higher-pixel density sensor, you need lenses capable of resolving higher.
     
  8. EarthQuake

    EarthQuake Mu-43 Top Veteran

    836
    Sep 30, 2013
    When comparing M43rds to FF lenses, you need to keep in mind that a M43rds lens at F2.8 "looks" like a FF lens at 5.6. I don't want to get too far into the equivalence stuff, but the DOF will be essentially the same. This means that, if an M43rds lens at 2.8 provides the look you want for a particular shot, a FF lens will at 5.6 (even taking into account boosting ISO, noise will be similar as well, though dynamic range may be a bit worse, especially on Canon FF) where it will typically be sharper. Consequently, M43rds lenses usually peak in sharpness around F4, while FF usually peak around F8.

    So yes, fast FF lenses are often noticeably softer wide open, but a wide open F1.4 FF lens is more like a F0.7 M43rds lenses, and there aren't any that exist to compare. The Voigtlander 0.95 lenses are similar to a FF F1.9 lens, and those lenses are soft ("dreamy" is usually the word used) wide open as well. Basically, FF lenses typically can be just as sharp as M43rds (and sharper on high res sensors like D810 etc) but they can also open up further if you need more light or want narrow DOF, at the expense of sharpness. Extra versatility, but you pay for that in size, weight, and generally, cost.

    I also do not think slrgear's ratings are designed to be used to compare lenses on different sensor sizes (source: http://www.slrgear.com/articles/interpret/interpret.html). DXOmark normalizes for sensor size and as far as I am aware is the only site that is theoretically comparable between different formats.

    When I shot FF, something I noticed over and over was that even with sharp lenses (my three favorite were my Sigma 35/1.4, 50/1.4 and 85/1.4, all of which were sharp wide open and exceptionally sharp at 2.8) the PDAF focusing sensors in DSLRs meant that you I rarely got a perfectly focused shot, usually slightly front or back focused, even after calibrating the lens. So I would usually stop down one or two stops to compensate for this and make the DOF a bit wider, and provide a buffer for the focus zone. This made my transition to M43rds quite easy, instead of always stopping down 1-2 stops, I can simply shoot wide open and trust the CDAF to be, more often then not, accurate. The only downside here was CDAF's propensity for focusing on distracting background elements and ruining focus entirely, rarely an issue on PDAF DSLRs.
     
    Last edited: Jun 28, 2015
  9. Clint

    Clint Mu-43 All-Pro

    Apr 22, 2013
    San Diego area, CA
    Clint
    If you had comparable prints from all of these lenses, what you think is not so good you might consider outstanding. The measurements don't really tell one how good of an image someone can get from it. Even under the right circumstances that Canon 50mm can make some very nice images. And the Canon 85mm f/1.2 and 24-70mm are considered two of the best lenses available. But pros probably bought the 85mm for specific uses which make the lens design shine, and that can't be viewed by charts and measurements. The Panasonic and Olympus lenses are part of the formula that make m4/3s sit on par with many larger cameras, so yes m4/3s lenses are that good.
     
    • Like Like x 1
  10. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    I think this is the answer: they established an arbitrary baseline different for each system with "reasonably close agreement".

    I did the same comparison with lp/mm value from lenstip but this values again vary for different pixel density and, if I got it right, cannot be easily converted through a "crop factor" (even if this article somehow suggests the opposite, probably assuming the same pixel density).
     
  11. Ulfric M Douglas

    Ulfric M Douglas Mu-43 Hall of Famer

    Mar 6, 2010
    Northumberland
    *cough*
    And there you have it.
     
  12. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    If I got it right from the slrgear explanation DxO software measurement are a lot different from the synthetic ratings that DxO publishes on their site. Synthetic ratings are what they are, not much more then a "five stars" rating system.
    I see no specific reason to expect for the "raw measurement" to be just "wrong" (unless their software is simply a bad software). I also think that resolution/sharpness measurements can be easier to do than the sensor performance that is often criticized on DxO.
     
  13. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    According to SLRGear, the Canon 24-70 f2.8 II @ f5.6 is about as good as the Panasonic 12-35 f2.8 @ f2.8, though the Panasonic is better in the corners. Though I'm still not sure quite what the SLRGear charts mean. Much like DXOMark, it's not a very transparent metric.

    With respect to DXOMark being the only site that is comparable between different formats, I heartily protest. Any site that publishes MTF values in LW/PH (Photozone.de, for instance) is already presenting measurements that are comparable between sensor sizes. Any site that publishes MTF values in LP/MM (Lenstip.com for instance) merely requires dividing the MTF value by the crop factor to make it comparable.

    These values are much more transparent and indepedently reproducible than DXOMark's P-Mpix values, which have no real published methodology behind them, just a "trust us, we know what we're doing," statement.
     
    • Agree Agree x 1
  14. Turbofrog

    Turbofrog Mu-43 Hall of Famer

    Mar 21, 2014
    For what it's worth, someone on DPReview did end up compiling those values, and here are the results:

    yVJTk2K.

    Ultimately, the very best FF lenses like the Sigma ARTs and the Otii probably end up reaching their peak resolution at f4 or f5.6, so this chart may not be perfectly fair to the full frame lenses. (Though a few of the M4/3 lenses like the 75mm/1.8 for instance, peak at f2.8, so maybe it is fair).

    As you can see, the M4/3 values divided by 2 are 37.5 @ center and 29 @ edge, compared to 41 @ center and 34 @ edge for FF. So there's a 10% improvement in the center, and a 17% improvement at the edges to going with FF. That's a linear measurement, so if you're looking at it based on area rather than the diagonal (which has a 1.94 crop factor) you end up with a 13% "megapixel" improvement in the center and a 29% "megapixel" improvement at the edges.
     
    • Like Like x 2
  15. tkbslc

    tkbslc Mu-43 Hall of Famer

    A "Perceptual Megapixel" almost sounds like something you'd see during a drug-induced moment of clarity.
     
    • Like Like x 1
    • Agree Agree x 1
  16. bikerhiker

    bikerhiker Mu-43 All-Pro

    Dec 24, 2013
    Canada
    David
    Photography is all about "blur", because 99% of it is blur. Only 1% is technically sharp. The question is; how does optical engineers design the lens to represent this 99% of the blur? Many had taken different approaches and produce a different aspect of image blur -- bokeh, resolving power and contrast all combined together to give you the representation of that 99% blur that you like. One of the Canon EOS Lens World workbooks illustrated this.

    So what is better and how do you measure blur?!? Well, you can measure it using instruments. In the end, it is all "SUBJECTIVE". A blur is a blur is a blur. It's really not sharp, so technically what we see as blur is acceptable sharpness. So what is acceptable? Again, another measurement that is also "SUBJECTIVE". There is no OBJECTIVE measurement on lenses like what DXO, SLRGear and DPreview are leading you to believe. In the past, Popular Photography used and still uses the SQF (Subjective Quality Factor) and acutance (Imatest) developed by the Kodak labs. But all of this is an attempt for them to lead you to believe this so that you can buy the best optical equipment that money can buy. If you agree with them, buy the lens.
    As an artist, I buy my lenses like I buy my paint brushes BASED ON what I want to create not what some arbitrary numbers derived by a machine. If I want to print big and ensure that if my clients were to stick their noses right in front of my 20x30 print, I better make sure I have the D800 or a Canon 5DS and Zeiss Otus take that photo with a sturdy Linhof tripod and no wind, cable release blah blah blah. But I know this because I know the requirements of my output result.

    How often do you do this? If your output device is a 2K display, then buy a good lens you like and you can afford. The majority of people I see with expensive pro lenses would never exploit the full resolution because the end result is always on a little teeny tiny iPhone, iPad or 2k display and maybe an occasional 8x10 or 11x14 print which pretty much all consumer level lenses can do. But it does make them feel better with their insecurity so own a Zeiss Otus or a Leica Summicron if you have to.

    Understand your output requirements, then buy the lens and camera to produce it and stay away from test charts and spec sheets.
     
    Last edited: Jun 30, 2015
    • Agree Agree x 1
  17. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    I'm not looking for the perfect lens. And I'm not a blind believer of tech reviews. I just noticed "inconsistent" chart results and tried to understand better.

    I think that some things can be measured and IMO it is worth the time to understand how to (correctly) read these numbers. How and how much these measures relate to the different aspects, technical or emotional, of a photograph as a whole is definitively "subjective" and more or less relevant.
     
  18. bikerhiker

    bikerhiker Mu-43 All-Pro

    Dec 24, 2013
    Canada
    David
    If blur is consistent, then it wouldn't be called a BLUR would it?
     
    Last edited: Jun 30, 2015
  19. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Lorenzo
    I think there is a misunderstanding. "Blur index" is the name chosen by slrgear for their own sharpness scale. "Blur" refers to the Photoshop command "blur". A blur index of 3 means that the picture is similar to a pixel-level perfectly sharp image after the "blur" command was applied three times. The lower the better.

    They shoot flat targets, there are no out of focus areas in their tests. Obviously no lens has a perfectly flat focus plane but out of focus has nothing to do with their "blur index".

    What in my opinion is "inconsistent" is that in the charts many of the m43 lenses show better sharpness performance ("blur index") than the Otus 55 that AFAIK is unanimously considered one or the best commercial lenses around (Sigma ART lenses are right back). So it looks like the Oly 75 is much better then the Otus and this IMO means that their measurements are "inconsistent" (as they are, they use different baselines for different systems).
     
  20. bikerhiker

    bikerhiker Mu-43 All-Pro

    Dec 24, 2013
    Canada
    David
    Ok I see what you mean. Remember that optically, a full frame lens and a m43 is the same. It's just the image circle of a full frame lens is larger than a m43 and because optically, the central portion of any lens has the highest optical performance (with little curvature), m43 lenses enjoy better optical performance because all the sensor see mostly is the center portion of the lens which then will register the highest resolving power. And which is also why shooting wide open @ f/2.8 or 1.4 does not suffer that much. @ 0.95, it's including the corners, which can drag the overall optical performance. A full frame sensor sees a much larger optical area than m43 and hence the performance of a full frame lens must take into account of the optical performance on the edges, which can reduce the optical performance of the overall lens. If you are to shoot full frame and do a 2x center crop of an Otus lens, then you can get similar optical performance like a m43 lens do because now you ignore the corner optical performance of the full frame lens which dragged down the overall number.

    If you look at the MTF chart, the graphs will reflect a nicer plot to cover the area of the m43 sensor. But full frame extends further to the edges and thus the resolving power continues to drop as it reach the edges of the frame. That's all, but really means nothing unless you have specific needs. Otus is designed for maximum resolution; ideally for high megapixel count cameras but our 12-40 and 40-150 are no slouch either.

    But to be honest. I have no clue what their Blur index meant just like I have no clue how DXO derived an 85 for their Super RAW on their new DXO One without much explanation either.

    Just treat numbers as numbers. If you need it buy it. I stopped trying to understand it.
     
    Last edited: Jun 30, 2015