A tale of three lenses - comparing the 50/1.4, 50/1.8 and 55/3.5 micro

Discussion in 'Nikon' started by kdghantous, May 11, 2010.

  1. This is a semi-formal image quality test of three old Nikkors: the 50/1.4 AI-S, 50/1.8 AI-S and pre-AI 55mm/3.5 Micro.

    Each lens was exposed at three apertures: wide-open, f/5.6 and f/16. Shutter speeds were within 1/3 stop. A Nikon D1X was set on manual with custom white balance. ISO was set at 125. File format was JPEG-fine at maximum resolution (approx. 5.7Mpx). No sharpening was applied at any stage. The same camera body was used for each lens. There is noticeable dust on the AA filter but this doesn't affect the results.

    The monitor was a BenQ 22" FP222W (soon to be replaced) and resolution was 1680 x 1050. The test target was a text file composed in TextEdit. Font was VT100 20pt. Text was a block of ampersands. Each ampersand measured 10 x 15px on the monitor. The character was chosen due to its shape. Although the VT100 ampersand is not ideal, the ampersand has a diagonal line, two differently sized loops and two short tails. The VT100 typeface was chosen as it was a bitmap font and could not be smoothed by the OS (one of the few problems I have with Mac OS).

    There is one problem with the 50/1.4: the diaphragm's sprung lever was not working. At f/1.4 it didn't matter. But at f/5.6 and f/16, I had to manually set the aperture with the lens off the camera, then re-mount it, being careful not to let the aperture lever on the lens touch the one in the camera. Focus had to be set wide-open and great care had to be taken not to allow the focussing ring to move as the lens was taken off and re-mounted. In hindsight I should have found a way to keep the DOF button depressed.

    With the 55mm lens mounted, the camera was placed slightly further away than were the other two from the monitor (in order to compensate for its slightly longer focal length). This was approximated only.

    I tried to get the monitor and camera as level and as aligned as reasonably possible. This should not be seen as a conclusive comparison, but I estimate that you can have 85% confidence of the quality of the results. Perhaps less with the 50/1.4, perhaps more with the other two.

    I expected the Micro, despite its age, to be very sharp wide-open, but it wasn't. The 1.4 was, as expected, poor wide-open, but I like it for portraits. The 1.8 looks worse wide-open than the 1.4. This is odd, considering I did check focus carefully before each shot, and a previous real-world test showed the 1.8 as a better lens in the corners. Perhaps my perception was wrong. So I did another test with each lens, wide-open only. The results were equivalent, therefore I didn't have to re-shoot the lenses wide-open. The reason for this might be that the 1.8 unit I have is an atypical sample.

    What seems to be light fall-off is probably the monitor's fault, not the lenses'.

    Just a note, though: photography has always been a mixture of technology and art. This is more than just incidental. Good technology is very important to photographers. That said, you can have a technically perfect lens which doesn't give pleasing images (e.g. unattractive bokeh) or a compromised lens (e.g. the Nikon 50/1.4 I tested) which can give pleasing results in portraits.

    Considering the cost of these older lenses, you don't have the right to expect Leica-like quality from them. If you want the quality of the Summilux M... buy a Summilux M. Don't expect to get optimal results with these lenses at anything wider than f/8. Otherwise a D2H may give you results no better than a D3X.

    This experiment didn't take too long. I actually performed it twice. The first results had to be thrown out as I wasn't 100% sure that focus was maintained for each shot. The second time around was easier and even more precise, so it was worth the trouble. In fact the second test was more satisfying and felt less of a chore.

    Everyone knows how much care can go into simple test. By doing one yourself you can experience it for yourself and really appreciate the work that goes into complex scientific experiments. This was just a lens test. It's hardly a big deal compared to what goes on in the sciences.

    I was surprised at the difference between the 1.8 and the 1.4. That taught me a lesson and brings the point forward that you should never assume that newer lenses - or 'slower' ones - are better.

    The results follow. Each image contains 1:1 cropped sections of the centre and top left corner of each shot.
  2. Next result:
  3. And the next one:
  4. The next lens:
  5. Keep on going:
  6. Next, please:
  7. Now it's the Micro's turn:
  8. I think I would have shot more apertures, especially f8, but I'll bet at f5.6, if you printed all three at 8 x 10 you'd see no difference. Did you do any print test?
  9. It's nice to see that the 1.4 wide open has an edge over the 1.8 wide open. It perhaps makes the 3X price tag worthwhile. Although these aren't the latest AF versions....
  10. Photographing a monitor may be as much a test of internal flare as anything else. Might be interesting to compare results with a more conventional test target.
  11. Thanks for the test! Subject to peer review, it DOES change long held ideas. I am surprised that the f 1.4 is as good or better than the f 1.8, which helps me to want a 1.4 with glee. I suspect that not all brands did their f 1.4 with equal quality, I suspect.
    The biggest shift in my thinking is that I was taught that some lenses actually start to lose some sharpness as they get to f 16, though increasing depth of field. Maybe joy begins at f/ 8?
    I'd like to see this confirmed with three different lenses. And, what about f/22? Great work!
  12. Diffraction is a different issue and on an APS-C camera, f22 is pretty useless. Here's an example at 100% pixels that I created a couple years ago when discussing this here. This particular lens sharpened up at f5.6 as sharp as it gets. Beyond F11 is clearly problematic.
  13. You shot a self-luminous subject, underexposed it, and posted the results as a test? Even a brick wall would be better than this. Take pictures, guy.
  14. For all three lenses, I would have expected f/5.6 to be near ideal and f/16 to suffer at least somewhat from diffraction. Yet in each case the f/16 image is sharper than the f/5.6. That would seem to suggest a focusing error.
  15. Focusing, especially with a dSLR, is a huge problem. IMO there will be systematic errors that cause the results to be sort of repeatable, falsely raising confidence in those results, but not saying as much about the lenses as might be hoped. For this sort of thing it's probably better to use a traditional 1951 USAF target or one of the newer designs made for digital sensors, rather than the monitor. Still, the results are interesting considering the popularity of the three lenses and it might be worth it to pursue more testing.
  16. I suspect that the focusing isn't perfect, since the f16 shots display more detail than the f5.6 and the fast 50 mm nikkors that I have used are really at their best at f5.6.
    One thing to note here is that the test subject has detail at basically two levels, the size of the letter and the size of the LCD grid. The latter is around what the sensor can resolve. But this then means that the capabilities of the lens are tested only at these two scales on a high-contrast target. In practice it is mroe useulf to have details at varying scales and lower contrast, since that's more likely to come up in real photographic situations.
    I should test my 50 mm lenses against each other someday when I have the time. I know which one is the winner though, but it's not the fastest one I have.
  17. I question using an LCD monitor as a test. As pointed out by posters above, the calibration charts are used for a reason. Then again, I'm not a fan of erastz lens tests, anyways. I'd rather be shooting something interesting. :D
  18. Results, while interesting, are a little against with my personal experience: on a D700, the 55 Micro 3.5 is already very sharp at 3.5, and peaking by F8 (but it's never "soft", unless I do make a focus mistake).
    A 50 1.8 AF is sharp by F2.8, and extremely sharp by F4.
    I wonder how much variability is present even in the old, nicely built gems...
    Funnily, my Nikkor Micro is beaten by a old Vivitar 55 2.8 macro, which sharp wide open too, and has generally sharper results with more pleasing bokeh... mysteries of old lenses :)
  19. What distances and which older 55mm F3.5, as there are at least two versions, one of which is a sharp macro but no good at all for distance? I have the non-compensating version and not sure I would use it except as a close in macro. The other normal lenses are general purpose at infinity and on in.
  20. Les, thanks... I guess!?
    Kent, I expected different results, too. I'm not sure about focusing errors, though, as the 1.4 wide-open would have been way off if that were the case. I have an idea about how to resolve the issue and I'll post a new thread next week.
    Jeffrey, I will try and acquire a Vivitar 55 macro. Sounds like one of those best kept secrets. ;-)
  21. Thanks for doing this :)

Share This Page