Friday trivia: 50mm shoot-out

Discussion in 'Nikon' started by fluppeteer, Aug 9, 2013.

  1. Hi all. Just a query. I'm just back from holiday, and this has been bugging me since I left.

    A few weeks ago, Amateur Photographer magazine in the UK did a comparative test of a number of 50mm lenses. I was very surprised to find the Canon EF 50mm f/1.8 II reported - if I'm reading their graphs correctly - as outresolving the Nikkor 50mm f/1.8 AF-S G even at wide apertures and, I believe, at the edges of the frame. I've always understood that the Canon nifty-fifty had the same formula (approximately) as the Nikkor AF-D version, and didn't think that it performed any better. I own both these lenses, though my camera bodies aren't really suited to a perfect comparison unless I resort to film, but AP's results seem to be, shall we say, inconsistent with my own experience of the two f/1.8 Nikkors and with other reviews.

    Does anyone have any suggestions as to what may have been throwing AP's measurements, or is my opinion of the Canon lens unfairly low? (I don't dispute that it's tack sharp stopped down, it's the claim of its sharpness wide open that surprised me.) Especially welcome if you read the article, obviously!

    Further information: I've just had a look at DxOmark's reviews of these lenses (on, arbitrarily, a 5D3 and D600). It claims that the Canon is very close to the AF-S wide open, and that by f/2.8 the AF-S has the weakest corners of the three (with the EF lens in the lead). This does not tally with my experience of these lenses, or with comments I've seen in this forum, that the AF-S lens is by far the strongest at wider apertures. I might guess that the Canon is applying stronger sharpening digitally, but otherwise I'm befuddled. Why might there be such disparity between the measurements and experience with these lenses?
     
  2. I might guess that the Canon is applying stronger sharpening digitally, ....​
    With DxO, shouldn't be the case; they work from RAW data only, if I am not mistaken?
    Frankly, the more I use the lenses I've got, the more my interest in the 'scientific' lens test sinks. It's useful to know a lens is not a total dog, but otherwise... I need user experiences (longer than "wow, awesome lens, it's sharp as a tack"). Testcharts miss a dimension, literally, but them another few dimensions too by figure of speech.
    I know most of my most prefered lenses do not come out on top with MTF charts, corner sharpness, field flatness and a raft of other aspects - and I know I am (probably) a bit over the top in my rather outspoken preferences for some of the gear I use. But in the end, it's really about the images a lens can draw. And some lenses do their own thing. To me, the 50 f/1.8G is just really nice in this respect; less sterile, more pleasant and alive in its rendering than the 50 f/1.8D. In focus to OoF transitions nicely, giving more depth to a scene; the f/1.8D is flat and lifeless in comparison. Colours reproduce nicer and richer too. There is no graph for that.
    Which is to say... a lot of the reputations of lenses (in my view) also boil down to these less measurable aspects of a lens (as a 200 f/2 owner, you got one hell of an example I think). I never used the Canon, so I cannot comment on it; but between the 50 f/1.8D and 50 f/1.8G, it's not just a matter of sharpness and such that sets them apart.
    (Not to mention I recently bought an AiS 50mm f/1.2 too... it doesn't disappoint in this fuzzy scale I use to judge my lenses.... it ends up high with beautiful funky glow-in-the-dark effects wide open).


    [edit] Hmmm. Hardly an answer to your question, is it? Let's chalk that up to friday being a friday....

     
  3. Oh boy, now you started it. Somebody will post a pic taken by a 50 1.8G just to justify that the magazine is wrong. :D
     
  4. To be fair, the 200 f/2 (although the new one, not the old and inferior one I own!) just got DxO's highest-ever 35mm resolution figures. :) But I agree about the unmeasured features, and one reason I wanted the AF-s lens was that the bokeh is much nicer than the AF-D version.

    Nonetheless, the actual thing that pushed me over the limit into getting the AF-s was that I was trying to check my shiny new D800 for AF point alignment issues, and the AF-D was so soft wide open that I couldn't tell what the edge focus points were doing. The AF-S was worlds better. I'm just surprised that it's not appearing that way in AP's or DxO's charts, and wondered whether there was anything to learn from the disparity - it might stop me making an expensive mistake with a future lens based on chart data, for example. I don't like things I can't explain. (I wondered if the default RAW profile might be different between the cameras - I guess it's possible that the 5D3 has a weaker AA filter than the D600, but the D800 results are similar. I'm assuming that the companies work on converted raw data; if they work directly with the raw file, that's different and probably depends on some profile information.) I suppose the AF-S would be softened by correcting its distortion...
     
  5. http://www.lenstip.com/308.4-Lens_review-Nikon_Nikkor_AF-S_50_mm_f_1.8G_Image_resolution.html
    http://www.lenstip.com/219.4-Lens_review-Nikon_Nikkor_AF_50_mm_f_1.8D_Image_resolution.html
    http://www.lenstip.com/27.4-Lens_review-Canon_EF_50_mm_f_1.8_II_Image_resolution.html
    These tests mostly seem to back up the AP report, regarding the D or AF-S. Maybe you have a really bad D or a really good AF-S? However, the Canon is not great wide open and doesn't improve so much on stopping down. That being said, it made great slides on my old EOS 5....until I broke it in an unfortunate digiscoping incident.
     
  6. Hmm. Thanks for the links, Mike. I've really been under the impression that my AF-S is pretty usable near wide open, and the AF-D isn't. Maybe I've not looked right into the corners, though. I thought I wasn't alone in my impressions, but maybe I am. :)

    This isn't stopping me waiting to see what the Zeiss 55mm can do - but then that's why I didn't spend f/1.4 money on a current 50mm already!
     
  7. I thought I wasn't alone in my impressions, but maybe I am. :)
    No you're not alone; we're at least in 2!
    No matter what all those figures try to tell me, on my cameras the f/1.8G is easily loads better than the f/1.8D, certainly below f/2.8. But I do not spend a lifetime studying the corners, either. But the newer lens just is nicer in each and every way. I find it usable wide open, and certainly near wide open (f/2.2 example, f/2 example), but I tend to like to watch photos not pixels, so maybe that's where I lose the plot ;-)
     
  8. Is there such a thing as a "bad" Double-Gauss lens?
    Perhaps, but I have both of the lenses mentioned above (Nikkor and Canon EF) and have found either of them nearly as good as the 1948 Zeiss Jena Biotar 58mm f/2 !
    (see Keppler's discussion of this point at http://keppler.popphoto.com/blog/2007/04/inside_straight.html#more )
    I'm sort of "funnin' ya" but these are venerable lens designs that are famous for their quality and, depending on what exactly is measured, I wouldn't be surprised at any of them floating to the top.
     
  9. Since changing over from Leica with Summicron 50 v4 to D800 I found that 50mm is quite a problem with Nikon, 1.8G seemed the way to go but I don't like it much, not in the same league as 85/1.8G. As 50mm-ish is a much used length for me I went looking for something better, my choice 45 PC-E and Voigtlander Nokton 58/1.4.
    Obviously neither are auto-focus, but for me the Voigtlander was a real find and has the kind of character that appeals to me.
    Sure PC-E is very nice too but you wouldn't want to carry it around all day.
     
  10. A few years ago I had a brief and unsatisfactory flirtation with Canon before reverting back to Nikon, and did a "shoot out" between the 50mm EFII and a 50mm f/1.8 AF Nikkor. The camera bodies used were a 5D and D700, so camera resolution was very similar. I did find that the Canon was slightly sharper wide open toward the edges, but the Nikkor had better central definition and a touch better contrast. However that sample of AF Nikkor was found to be inferior to my old Ai-S 50mm f/1.8 and throws into doubt whether those results might be typical or down to sample variation.
    Stopped down a bit there's very little to choose between any of those lenses - except - what I do hate about the Canon lens is its obvious barrel distortion, which I find pretty unforgiveable and unacceptable in a 50mm lens.
    PS. Mike, why would those lenstip tests show the same lens to have a different central resoltion on FF and DX? Something smells fishy there!
     
  11. Addendum. I've just revisited the lens tests I did when I first got my D800; mainly to find out what was fully useable on the camera and what wasn't.
    My tests of the old 50mm f/1.8 AF Nikkor show that it's not a lens that's going to shine when pointed at a resolution chart. It appears to have an "S" shaped curvature of field that takes the resolution down considerably toward the edges of the (full) frame. Surprisingly the field comes almost back into line with the centre at the extreme frame corners by f/4. Annular zones of good resolution can be selected simply by tweaking the focus. Naturally this field curvature will show as poor definition on a 2-dimensional test chart, but in real world pictures it probably isn't very significant.
    By f/2.8 in the centre of the frame the lens was resolving about all that the D800 could manage, and didn't improve significantly on stopping down below that. Although overall contrast did continue to improve slightly up to f/5.6, the extremes of the frame didn't really ever catch up to the central definition. What holds the edges back at wider apertures is coma smearing, spherical aberration "haloeing" and a general lack of contrast. The actual resolving power remains quite good though, and I suspect that some smart software to increase localised contrast could make this lens look several hundred percent better.
     
  12. Isn't comparing the sharpness of 50/1.8 lenses splitting hairs? They differentiate themselves on contrast, flare control, CA
    and bokeh, not sharpness.
     
  13. Just to add some reviews that that I think to offer quite balanced conclusions and include samples you can look at full resolution (even considering the net limitations).
    The Canon:
    http://www.photozone.de/canon_eos_ff/415-canon_50_18_ff?start=2
    The Nikon AF-D
    http://www.photozone.de/nikon_ff/623-nikkorafd5018ff?start=2
    The Nikon AF-S
    http://www.photozone.de/nikon_ff/631-nikkorafs5018ff?start=2
    At the end of the day there are always trade-offs when it comes to a buying decision.
     
  14. Andrew,
    I repeated your experiment with DXO using the same cameras as you and confirm the result; the corners of the 1.8G just never come good. www.photozone.de shows the corners of the 1.8G to far better than those of the 1.8D at all stops wider than f/8. My guess is that DXO messed up or they had a defective 1.8G. Don't know about AP though. As expected, on www.photozone.de the Canon looks better than the 1.8D and not so good as the 1.8G.
    My experience of the 1.8 AIS (compact model, same design as the AF I believe) on film and on my D7000 is that it's poor wide open. On the D7000 my new 35mm f/1.8 DX blows it away wide open - quite a bargain at £145.
     
  15. In my experience, the 50/1.8G on the D800 has great overall sharpness, and if there's any problem with the corners it's
    not noticeable. If DXO comes to the conclusion that it's not a sharp lens it's too nitpicky and this discussion is silly.
     
  16. "Isn't comparing the sharpness of 50/1.8 lenses splitting hairs?" - Well, sharpness encompasses contrast and CA, not just resolution, but I don't think this thread is just about sharpness anyway.
    What I think is interesting, and I take also to be at the core of Andrew's question, is how a well-established basic design like the double-Gauss or Planar can be interpreted by different companies and their optical designers. As you say Andy, it's all of the factors that go into a "good" lens that give you an insight into the broader design philosophy of an optical company.
    Nikon, for example, always seemed to have prized central image sharpness, good contrast, control of CA and flare above field flatness, distortion and corner sharpness. This played well in the days of film, with its ability to capture an image over a shallow range of planes, but with the absolutely flat and 2-dimensional image plane of digital, some of their old designs were found wanting.
    I think Nikon's optical designers are still coming to terms with digital. Even more so having just basically shot themselves in the foot with the extreme resolving power of the D800. So their tweaks to a donkey's years old design of standard lens is a pretty good barometer of which values they hold dear.
     
  17. Maybe, just maybe, the apparent disparity between some user's and reviewer's opinions suggests that there's more sample variation on this 'slow, consumer-level nifty fifty' than we're used to....??
     
  18. I own a bunch of lenses in the 45mm to 60mm range. They all are good for very slightly different things. For sharpness and all around image quality the best of them wide open is my old factory converted non AI 50mm f/2.0. It is more then sharp enough wide open and by f/3.5 is as sharp as anybody could ask for. After that it is a very close three way tie between my 50mm f/1.2 55mm f/1.2 and 50mm f/1.8 G. Stop any of them down to 5.6 and who really cares which one you pick up they are all very good.
    I have never been a big one for using strictly controlled testing as the only means to judge the quality's of a lens. resolution is only one of the things I look for. I am also concerned with the way a lens "draws" an image.
    I do have one lens that I find to be extremely sharp wide open and it is in the 55mm range. But it is not very useful for general photography as it has no focusing helical. But if you ever get the chance to play with a Nikkor O 55mm f/1.2 you will be amazed at what it can do.
    http://www.flickr.com/photos/14133178@N04/7235682302/
    The difference between Cannon and Nikon 50's? I would think is not enough to really worry about.
     
  19. I have the Canon 50 f1.8 and the Nikkor 50 f1.8D and G lenses. I've never tested them since my Canon body is only 6mp compared to 12mp of my D2x and D300s but to my eyes both Nikkors look sharper.
     
  20. Sorry not to have checked back over the weekend... thanks for the feedback, everyone. To clarify, the reviewers weren't critical of any of these lenses, and I do have all three (though my only digital Canon is also a 6MP crop body) so I'm not inquiring for shopping reasons. I've not spent time shooting test charts to confirm, but I just wondered whether the results (expressed in numbers rather than text) were expected, and was trying to understand them - not because I need to know for these lenses, but because I wanted to explain the surprise.

    To be honest, I've never been that impressed (in technical terms - I appreciate the charm of undercorrected spherical aberration) by any 50mm at wide apertures (which is why I didn't spend more on an f/1.4); I do expect all of them, especially the double-Gauss derivatives, to be extremely sharp when stopped down - there are no terrible 50mm lenses - but since higher pixel densities mean that diffraction is starting to be significant at larger apertures these days, I'm paying more attention to how lenses perform at f/4 and larger. (Besides, it means I spend less time dust spotting.) That said, on my D700 I spent more time wide open, whereas on my D800E I tend to stop down more to maximize sharpness. I'll be interested in what the new Zeiss can do when it's launched, but keeping an f/1.4 manual focus lens locked on a moving subject is likely to be hard enough that I'm not sure how many of my subjects it'll help.

    Thanks for the additional evidence, anyway. I must get a book on lens design.
     

Share This Page