SEL20F28 impresses and puzzles at the same time

Started Apr 10, 2013 | Discussions thread
OP blue_skies Forum Pro • Posts: 11,470
Re: now, now ...

verybiglebowski wrote:

Annex wrote:

007peter wrote:

Off course Software Correction is cheating.  Because we are comparing a:

  • Non-Software Corrected Sigma 19 vs
  • a Software Corrected Sony 20

A better design fair test would be like these:

  • A software corrected Sigma 19 vs
  • a software corrected Sony 20 or even
  • a non-software corrected Sigma 19 vs
  • a non-software corrected Sony 20

the problem here is that if I go by your [viking79] reasoning, I might as well claim that my

  • $300 Canon 17-85 software corrected via PTlens is
  • Much Sharper than any $1200 Canon 17-55

No it isn't - Some lenses are designed with Software correction specifically required (eg the 16-50). Others are not. If you compare an old lens (pre software correction) with a modern lens designed for it to be applied you are artifically biasing the test. If the artifacts are not automatically correctable in standard software (in camera or popular image suites), then it's genuine, manual post processing and I agree that is a different case.

However to review or critique lenses that are quite obviously only supposed to be used corrected is just silly. Its a denial of progress.

Eg the 16-50 is approx 14mm at wide end in raw with major vignette, distortion etc. It's intentionally like this as it's done to keep the frame sharp in the full knowledge that vignette and distortion are fairly simple to correct for automatically but that sharpness is not. To test this lens without correction missed the point and would restrict lens design artificially. From what little I've seen so far this is also the case for the 20.

What matters is the end photos. Not how you get there.

Sure, but what are you comparing then? Post processing abilities, in camera engine, what? You can't compare lenses that way, you can only compare systems. If you say, that only final image matter (and I do agree), than let's discuss final images and not those useless test shots... And if we want to compare lenses, than let's do so using the same methodology - same correction procedure, starting from RAW files, MF etc.

Now, now,

I'd be happy to remove such useless test shots, just let me know. I have a lot more of such useless test shots - different scenes, different lenses, different lighting. These are all useless by your standard, but very telling to me as to what to expect from a lens.

And with different sensors. Especially the WA lenses, as they do perform differently on different sensors, most notably the 14Mp Nex-5, the 16Mp Nex-5N/6 and the 24Mp Nex-7.

In past, I have tried to get the 'best' picture possible, using RAW, JPG, different post-edit SW packages, and have made my own conclusions: some lenses work better than others, consistently.

To be more specific: wide open apertures (at or below f/2.8) show soft far corners on almost all lenses at 28mm or wider. (The S19 and E24Z are notable exceptions.) Especially on the 24Mp and 14Mp sensors. Stopped down, most clean up, but not all. And on some lenses it is field curvature related - and is most prominent at certain distances to subject (at least in comparisons).

Other than the brick or newspaper on flat wall tests, most photographers just know what to expect from a certain lens and when to use it, or better, at what FL and Aperture to not use a lens.

As to the test, yes, the AF focus shift was disappointing and unexpected. But otherwise, except for the CV21, these are all E-mount lenses, with in-camera corrections. Same engine, same 'processing flow', tripod-based same framing, and still a valid approach in my book.

The E20 is getting remarkably sharp at f/8.0. It is getting there at f/5.6, but the far corners may show softening (depending on image framing), which worsens down to f/2.8 (on both Nex-6 and Nex-7). I think that it is field curvature and/or correction related, because the Sigma 19 does not show any softening at all in the comparison shots, and neither does the E1650P (up to f/4).

I derived a number of conclusions from my experiment, and thought that sharing would help others in forming opinions. Again, this is non-scientific, but I have a lot more images than I can post here, (I already hit the upload limit), and these tell me that the E20 will not my preferred lens for landscapes. I would rather use the S19 or the E1650P lens - it shows in my pictures, focus point shift or not, and I will still keep using the E16+ECU1 adapter.

But it also tells me that for social settings, the E20 is quite usable, more than the E16 in fact. Typically, in such shots, the corner/edge performance is not relevant, and the speed (f/2.8) is kind of ok. I have used both the S19 and the E24Z in such conditions, and did not like the f/2.8 too much, but the associated DOF increase makes up for it, versus e.g. f/2.0 on the E24Z. Like others have pointed out - the Nex with E16 or E20 looks like a P&S, and I get more smiley faces with it than with the S19, which makes the Nex look more like a camera. Forget a DSLR - it comes across as being too serious.

Besides, I don't think that any pure scientific testing will lead to different conclusions. The S19 is remarkably sharp at all apertures, and so is the E1650P, at 20 and 30mm. The E24Z also produces remarkable results at all apertures. The E20 struggles, especially wide open, not unlike the E16. Some love the E16, others hate it. I expect a similar fate for the E20 to develop over time, it is better than the E16, but not perfect. (It never bothered me - you can frame differently, stop down, or crop in post.)

I don't know what the Sony engineers are dealing with - or how much it would take to improve such lenses. My take is that it takes a lot more size - the E16, E20 and E1650P are remarkably tiny little lenses. Professional grade? Maybe not. But for amateur vacation snaps remarkably useful in a tiny APS-C package. If the E16 and E20 would have been f/4.0, would it have been better? I am glad that they are f/2.8 - but the camera software should only use f/2.8 as a last resort.

I took the S19 and S30 (only) with me on travels before, as smallest and lightest option (in a tiny bag). But I eventually switched back to the E24Z and E1855 lens, for several reasons. Now I may just take the E20, E1650P and E35. Better would be if I only picked two, and not three lenses.

Locally, a small camera bag which holds 5 or 6 lenses gives a lot more flexibility. I would not take the E20 with me then - other lenses would be preferred. (I would probably still take the E16+ECU1).

Lastly, for my type of 'social' snaps, keepers are not defined by lens or edge performance - rather, they are defined by the subject being in focus, in a flattering pose, and with a great smile. The (far) corners seldom matter when people are in the picture, unless in front of some scenic background, etc.

One thought for Sony engineers - allow an auto-crop on the E16 and E20 lenses - e.g. crop the E16 to 18mm and the E20 to 22mm - it will possibly hide all (far) corner artifacts, and make the lens a lot more pleasing to critical reviewers. (This can also be a use-mode for 'critical' shots).

-- hide signature --


 blue_skies's gear list:blue_skies's gear list
Sony Alpha NEX-6 Sony a6000 Sony a5100 Sony Alpha a7 II Sony Alpha a7R II +36 more
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow