Wednesday, 9 November 2016

How to get the best out of the Sony 16-50 OSS Pancake - A Contrarian View




So this folks is a test image taken with the Sony 16-50mm Pancake 36mm @ f6.3 at my standard tests location this represents the longest focal length before image quality starts to decline, it is still more than useable. Check out the 100% bottom left corner crop below, not too bad I would say.


Next time you are on a photographic forum and you feel like throwing the cat in amongst the pigeons just casually drop into the conversation that you think the Sony 16 to 50mm pancake is "not actually too bad". I can almost guarantee that you will become an instant pariah, targeted for unmitigated abuse and your credibility, well you won't have any amongst the photography cognoscenti from that point on.

But is the diminutive 16 to 50 mm OSS pancake really such a horror or does it in fact have some redeeming attributes?  let's discuss that.

Even the most ardent critique would have to concede that the lens is nicely compact, quite well made and a perfect physical match for some of the smaller A series “E mount” cameras (yes I know it is all very confusing now).   Of course they then couldn't helping adding the lens is completely fuzzy and little better than the bottom of a dirty beer glass. 


Well I’m here to tell you it is much better than most forum lurking critics believe, I can say that because I have actually tested the little blighter on a real world subjects using a structured approach. So here's the good bit, what I have ascertained from this testing may actually of be some help to those who also use this little package for real world pics, rather than shooting the obligatory test charts and brick walls, hence this article.

Yes, of course it is far from perfect, but an “almost free” kit lens can never be expected to perform anywhere near as well as an expensive fixed focal length lens, some perspective is needed, and let’s not forget most shots only end up as web images or maybe small prints anyway. In any case there are zillions of these little lenses out there attached to Sonys' and it would be a pity for people not to at least get their moneys worth out of them.





100% Corner Crop @36mm f6.3, Note and this is very important, all the pics in this article are RAW files processed in Iridient Developer, no jpegs allowed if you really want to test the mettle of a lens.


Special Note: Please accept that the actual images probably look quite a bit better than what you see here, fact is, the web compression and the device viewing conditions will always negatively effect what you see in the blog, if a lens produces rubbish I will say so quite happily, but likewise if I say it looks good on my 27 inch 5K Mac Retina screen at 100% view you can be quite sure it is.


The Need for Compact Option

My own initial interest was triggered by my desire to have a compact walk around camera that I could slip into my man bag as I wandered around the streets and byways of foreign cities. I've looked at all sorts of options, Sony RX 100 series, (lovely too but limited in application for me), Olympus and Panasonic mirrorless bodies with pancake lenses ( great but I’d then end up then investing in a whole new system, (which is ultimately what I did, but that is another story), Nikon new DL 1 series compacts, (not yet a known quantity, does they even really exist?) and so on.

Funnily enough I had early on discarded the idea of buying Sony A6000 or 6300 to replace my now well worn NEX 5n travel camera because I had believed that the kit pancake must have been some sort of “evil” and no other Sony lens option/combo would be compact enough for the task I had in mind.  Yes I know about the pancake 16 and 20mm options but they don't suit my needs. It’s not like I don’t know the Sony system well, I have used the Sony Nex 5n for nearly 5 years and have a wide array of legacy lenses and DIY’d bits for it and many A mount lenses and a couple of bodies.

After initially almost biting the bullet and buying an Olympus om-d-em10 mk2 with kit pancake I though why not take a pause and test one of those 16-50s and see if it really is a complete stinker!   If it was Ok then a nice shiny new A 6000 or 6300 could be on the potential menu, I'm glad I did the test and maybe you will be glad also.

One criticism of lens tests that often make is that "testers usually only test JPEG output images", now consider this, there is a radical difference between the quality of JPEG "cooked in camera" images and those that can be liberated by using a state-of-the-art raw converter, and this is especially so with Sony cameras.


In Camera Processing


Lenses with low levels of micro contrast often have their overall quality compromised by the native noise reduction processes which take place within typical "in camera JPG processing" methods, often turning fine details into, well let's say..... watercolour mush. 

It just happens that most kit lenses and many of the cheaper zooms don't have high levels of micro contrast and often once files have been run through the standard camera processing and noise reduction protocols they end up looking far less crisp than equivalent images from much more expensive high micro contrast lenses, I guess you could say high end lenses have more “sharpness wiggle room”.

In the case of the Sony 16 to 50 the problem is somewhat exacerbated by the fact that Sony's implementation of noise reduction has tendencies to smooth over fine textual details anyway, this is intended to keep apparent noise levels low, something which it actually excels at. But..... a side effect of this NR is that for many lenses the JPEG renders a far inferior result to that which could be achieved through appropriate RAW processing. Softening affects are further exacerbated when the ISO is placed above 800 mark, which is a common occurrence for the average shooter, especially when shooting in full auto modes with slow kit lenses.

Sonys’ JPEG processing has certainly improved over the last few years but to my eyes it still has a significant tendency to confuse fine detail with noise, additionally Sony's sharpening algorithms are adaptive but they tend to sharpen medium and high contrast details rather than very low contrast details, again I imagine because Sony wish to keep the apparent noise level low, this makes sense of course as noise reduction and image sharpening are actually flipsides of exactly the same coin.

Bottom line is this, the 16 to 50mm pancake performs considerably better when shot in the raw format, in fact I would go so far as to say it's almost like a chalk and cheese difference. 

If you did nothing other than switch your camera over to RAW and learnt how to use a good quality RAW conversion program you will radically improved the capability of the 16 to 50 mm pancake, but there’s more, a lot more that we can do.




28mm @f6.3, no problems here, check out the edge crop below, taken off the right hand side.





28mm Edge Crop


Not for Brick Walls


The 15 to 50mm pancake will not thrill the thriving “Masonry Wall Photo Collective Group” out there, it’s certainly not optimal for is capturing “flat field” objects in other words being used as a macro or "close to macro" lens, in this realm it is really quite useless.  I seriously doubt that Sony had any expectations of the average user buying the 16-50 pancake to shoot high quality macros and flat art. Mind you macro work of the 3D kind will probably be fine, just don’t expect good results on stamps and coins etc, or of course that lovely brick wall up around the corner.

The 1650 pancake in fact it exhibits considerable field curvature at almost every focal length, which means testing it on a flat chart, as most lens testers are want to do, would and does make it look pretty bad, though in fact for normal 3 dimensional medium to distant photography applications it's fine. (I have a blog post on Field Curvature and how to use it creatively if you are interested, you can find it here: 

https://braddlesphotoblurb.blogspot.com.au/2016/10/decoder-field-curvature.html

Briefly, field having curvature means that when the centre is in focus the plane of focus is not flat, it may curve away from you or back towards you as you move out towards the edge and corners of the frame. In practice a lens with field curvature generally gives a sharper overall focus if the focus point is placed around half way out from the centre of the image towards one of the corners, which corner will depend on the scene and type of curvature present. Many lenses with soft corners actually come reasonably good when focused thus.

Most cameras are factory set to focus using some version of “wide area” focus where the camera chooses where it thinks it should focus and despite the sometimes complex algorithms involved, this method can be somewhat random and I feel may account for why many folk claim the shots from the 16-50 are random in terms of overall focus. Basically this is the deal, quite often the auto focus point selection works against the field curvature and sometimes with it and occasionally it takes a middle ground, net result...... variable focus quality, which of course is a complaint many folk make about this little lens.

In all of the initial tests shots I made I used manual focus, setting the focus point around 30% in towards the centre from the bottom right hand corner. 

Note: This may not be the ideal point but it served as a pretty solid starting point, and generally I found the focus across the whole frame was quite good, regardless of how close the subject I was shooting was. I also tried some shots with the focus point in the middle of the frame and in almost all cases the overall cross-frame clarity was worse. I must point out that this is not an issue specific to the 16-50mm pancake, almost all kit lenses and a great many fixed focal length lenses I have tested show similar characteristics.

Wide area focus may be a wise choice for close to medium distance photos or where you are aiming to isolate a specific subject from the background, after all in these cases who cares about soft corners etc but it is not a wise choice for landscapes and photos where you have objects that range from the medium to infinity distances. For the record it looks to me that the 16-50 pancake set at 16mm has field curvature that moves the focus point further away from the camera as you move towards the corners, meaning if you focus in the centre of the frame the nearest details in the bottom left/right corners will be quite soft, bringing the focus back closer towards the camera leads to a better overall average clarity. This is worth checking out on your copy as most folk complain the 16mm setting is rubbish, something I now know for sure is not strictly true.


Flares Are Not In

You want more, there is much more…..the micro flare control of the lens, especially at the widest apertures and in the 16 to 21 mm range is not ideal, overexposure will lead the highlights bleeding into the surrounding dark pixels. The simple expedient of paying careful attention to your highlight rendering and controlling the exposure often by just 1/3-1/2 stop can lead to better overall images, once they have been edited to taste of course. 

By f5.6 the flare problem has disappeared at all focal lengths, (except 50mm, but even then it's pretty minimal) so if you are going to shoot at the wider apertures certainly pay a little extra attention to exposure.




This is the much derided 16mm setting, without distortion correction applied, not too bad eh, what if I was to say it was shot at f3.5 wide open, would that be a bit more impressive, well yes it was shot wide open! Yes, yes I know there is vignetting but we are talking about resolution at the moment, we I'll get to that soon.





16mm f3.5 Centre Crop 100%


Best Focal Lengths

One enormous shortcoming of most zoom lens tests is that testers usually only check the lens at the widest and most telephoto settings with a further test set somewhere in the mid range. In the case of the 16-50 pancake I could only uncover tests taken at 16, 35 and 50mm despite some exhaustive searching.  

Now listen here, fact is, very few lenses ever work best at the extremes and who is to say that 35mm is the sweet spot anyway? 

Most kit zooms have a sweet spot somewhere in the focal length range, a point where chromatic aberration and distortion are minimal and most likely cross frame clarity best. Some have a fairly wide sweet spot, which is great and some have a pretty tight envelope of reasonableness.

So what about the 16-50? Well there is plenty of good news! The 18mm setting is better than the old 18-55 OSS kit lens, quite a bit better in fact. 19mm proved to be actually about as good as the Sigma 19mm f2.8 so long as you were shooting at say f5 or smaller!

The 21mm setting seemed to be particularly good and I would rate 24-26mm as ideal, with 28mm was almost as good. In fact I compared the 28mm setting to the 30mm f2.8 Sigma and once the files were adjusted for optimal colour/tone/detail rendering the differences were very small, indeed I would rate the pancake as a tad better in the very centre of the frame. By 35mm the quality is starting to quickly decline but at f6.3 it’s perfectly OK for most needs.

The 50mm setting was easily the worst and I would use that only if I had to. The 50mm setting on the 18-55mm OSS was much better overall, though neither are a match for a good fixed 50, of course Sony offers a very nice 50mm f1.8 which could pair nicely with the pancake or you could opt for the Sigma 60mm f2.8, a real standout budget lens, though it lacks OSS of course. 

For the record, just in case you were wondering, the old 18-55mm OSS falls off the cliff at 55mm and I never use that setting, rather I shoot it at 50mm and crop later to get the 55mm view.

In summary used between 18-30mm the 16-50 pancake is really quite a good lens, providing of course you are not shooting wide open. I must add however that in the 21-28mm range you could in fact shoot wide open and the results are still really quite fine. Lets put it this way, for any normal usage I would be happy to go wide open at these settings, the main trade off will be bleeding of highlights, keep the exposure in check and all will be hunky dory.

And now some bad news, well not really bad, but worth knowing. All cheap lenses seem to exhibit some degree of de-centering, which in practice means one side or corner may be softer than the other/s. The 16-50 mm is no exception, the one I tested displayed this characteristic, mainly at 16mm and 50mm. Was it extreme, no, not at all but at a 100% on screen view you could clearly see it, stopping down to f5.6-8 eliminated it. You could be lucky and get a perfect example of the breed but I doubt it, if it is minor and this one was, shifting the focus point around may kill the issue, I tried that on a subsequent set of images at 16mm (once I had ascertained what I was dealing with), and got excellent results. 

Note, you could get a copy that is really out of whack so decentering is something to check for in any cheap kit lens.


OSS Oddity?

Now for something completely different! When looking very closely at the files at a 100% I came across an anomaly that seemed to effect every focal length, shots taken at 1/640 sec appeared to have unsharp areas in them which were somewhat random. What was weird was that these frames were less sharp than the preceding frames which was shot at say f5.6 @ 1/800 or the following frames shot at say f7.1@ 1/500! 

In other words my friends there was an unexpected resolution hole. Odd eh, then I had an epiphany, was this caused by an interaction between the OSS and the shutter speed, was this a shutter speed where the response time of the OSS and the shutter speed just clashed in some strange way?

Easy to test, go back take some shots at 1/640 sec with and without OSS, and guess what, yep I was right the ones without OSS did not exhibit the issue! My curiosity got the better of me and I re-checked the shots and found a similar issue but at a much lower level at the 1/1600 sec mark. 

Please note, this test was done on a Sony Nex 5n, I didn’t get to check later on the A6000 I had handy, so it might be camera model specific, nonetheless it is worth knowing and checking for on your camera. 

This OSS interaction may also account for some of the patchy results many critics claim to see from the lens, after all 1/640 sec is not an uncommon shutter speed to find your camera shooting at during everyday outdoor usage. Of course the problem is easily resolved, turn off the OSS if shooting in bright daylight, on most Sony models you can assign that option to a custom key should you wish.




18mm at f6.3, this is a critically sharp focal length and is better by a fair margin than 18mm on the older 15-55mm OSS lens.




18mm f6.3 Centre Crop @100%, bet you cannot even see that person in the middle in the full frame above!


The Best Apertures

If you’re looking for the very best results sharpness wise, having some idea of the optimum aperture for any given focal length can be helpful, most reviews never look at this in detail usually picking to test at wide open f5.6 and f8 and maybe 11. But what if the truth of the matter lay somewhere in between these? Having checked it all I reckon I can give a pretty good summation for you. Here goes:


16mm f7.1 (stop there, going smaller definitely causes diffraction issues)

18mm f5.6

21mm f5.6

24mm f5.6

30mm f7.1

37mm f8

50mm f10


Now all of the above settings assume you are nailing the focus and not trying to mask poor focus by stopping down more than ideal.


Focus Accuracy

To a degree the accuracy of the auto focus is largely camera dependent and certainly in testing this lens on both an NEX 5N and the much newer Sony A6000 it was obvious the later model camera is far more adept at nailing a moderately precise focus. Most lens reviewers only review using auto focus but let me share a secret with you, even though auto focus is generally pretty good you can often get a more fine tuned focus by using manual, especially with cheap kit lenses. In the case of this lens for the second set of tests I had the focus in DMF and then fine tuned the focus once the focus confirmation beep went off.

Almost without exception I could get a slightly better focus using this method than the cameras auto focus alone would obtain, and remember that this is in good daylight, so the difference would certainly be more exaggerated as the light level dropped, I put the need to do this down to the lenses lowish micro contrast making it hard for the system to see the maximal point of clarity.

Is the 16-50 particularly bad in respect to this? I don’t think so but I am sure it would not be anywhere near as good as a prime or high end zoom, of course there has to be some reason for spending the extra money. The take away being this, if you have the time and want to get the most out of the lens, use the DMF function as it will likely give you a sharper pic under a greater array of situations.


Chromatic Aberration

These days reviewers and photographers will commonly repeat the now accepted refrain that Chromatic Aberration doesn’t matter as you can easily fix it in software, true to a degree, you can fix it and of course the camera does so internally for the jpegs.  


CA is more serious than just being an annoyance, having quite a negative effect on edge and corner clarity. Most of the clarity losses can be reduced, provided the software is able to do a really good job, fact is however there are some significant differences between the results from different RAW applications, but all of them seem to do a better job than Sonys’ “in camera” processing. My tests seem to indicate Sony cameras are not particularly good at correcting CA without trading off some corner and edge detail so again RAW is best.

With the 16-50mm pancake the uncorrected CA is not actually that terrible, I have examined many uncorrected RAW files and overall I think it is actually quite a bit better than many of the kit lenses I have analysed, including the Sony 18-55mm (at least over most of the focal length range).

As expected it's worst at the widest settings and seems to be most minimal at 30mm and longer, in fact you could probably not even bother fixing it at the 30mm setting unless you were printing big. The 50mm has the lowest level of CA prior to fixing, in fact even at a 100% view it is very hard to see it in the uncorrected file.




24mmf6.3 Edge crop @100%





24mm @f6.3, sharp from edge to edge and corner to corner.


Distortion

Freedom from distortion is in not the 16-50s strong suite, with many lenses it's something that often can be just left alone unless you are shooting architecture and occasionally it's a good idea to do just that as correcting distortion will often degrade edge and corner clarity.

Unfortunately this lens really needs the distortion attended to when shot in the 16-21mm range, the 16mm length in particular being occasionally quite unsettling. There is however one giant fly in the ointment, the standard corrections method used in camera and by default in RAW conversion programs significantly crops the image, in fact I would say the lens is actually more like a 14.5 mm one if left un-corrected.

Cropping itself is no major drama but in this case at the 16 to 18mm settings the “de-distorting process” radically messes up the edge and corner clarity. When you turn off the corrections in Raw you quickly realise that not only is the 16mm much wider than stated but the outer resolution is potentially quite good, especially if you do a custom CA correction.

If there are no straight lines near the edges of the frame you could easily leave the 21mm images uncorrected, 24mm I would not bother except for architecture, 30mm could generally be left alone, same for 37mm. The 50mm has a slight amount of pin cushion distortion but you would almost never detect it. My guess is that there’s probably zero distortion somewhere around the 40mm mark, which could make this a good focal length to choose when creating stitched images.

For images shot at 16mm you will get significantly better results by fixing the distortion in Photoshop or a similar app by using the advanced transform tools, as said, it is quite surprising how much extra image area you will liberate via this method.


Vignetting

Once aspect that will show up if you stick with the “full wide” 16mm rendition is that the very outermost corners are very heavily vignetted, that might be occasionally problematic but in many instances such wide angle shots end up in 16:9 ratio crops which cut off the outermost corners thus you would never see it, if you really want the corner vignetting corrected you could make a custom edit in Photoshop using levels and layers or perhaps apply a very selective version of the vignette correction tool, often you can easily clone the corners if they have sky or cloud in them.

Ultimately when shooting at 16mm manual distortion correction will give you much sharper images overall, trust me the "in camera correction" really does seriously mess up the edge and corner clarity in the16mm-18mm range, you might even like to try shooting your jpegs with the correction option turned off when not really needed, you may just be a pleasantly surprised.


The Wrap Up

So to conclude, would I use this as a walk around general purpose lens, hell yeah, no hesitation at all, now that I know my way around its characteristics. Further more I reckon lots of Sony users will be able to get really nice results from it, it’s no silk purse of course but it is much better than a sours ear.

The only real reason for reaching for another lens for regular snaps and holiday stuff would be if you need a more shallow depth of field rendition or a longer focal length, or perhaps if you often shoot pics at the 50mm setting.

It is actually a fun little lens, I quite like the way to zoom works and love that it’s so compact and adds almost no weight to the camera, it also works a treat in combination with the DMF (Direct manual Focus) option. Add into the equation that the 16-50 is crazy cheap to buy with your Sony camera at purchase time, it would make little sense to not grab the camera/lens combo.

If you are prepared to work around its foibles the results are potentially very good, what more could you ask for when so little money is involved?


The Good

Very light and compact

DMF works well

Electronic zoom is good for video and it’s silent.

Has a reasonably wide sweet spot of clarity

Cheap

Nicely wide when you need it.

Even wider is you turn the lens correction off.

Suits a lot of general photography needs

Generally sharp enough.


The Bad

Distortion at the wide end is fairly extreme

Serious vignetting at the wide end of the range in RAW

The longest 10mm of the focal length range are pretty average with 50mm not much good at all, however it's fine for portraits when you are after clarity in the central image portion only.

Slow of aperture, but then no different to other kit lenses

Suffers from significant field curvature, especially so in the sub 21mm range.

Not really suited to macro duties but passable with a plus 2 diopter close up filter and no worse most other kit lenses

No lens hood supplied, it is not significantly flare prone but the addition of a hood does help a bit.

Low micro contrast.



What You Need to Know.

You will get vastly better results with this lens if you shoot in RAW and edit with a state of the art RAW editor, the results from Iridient Developer and Raw Therapee are vastly better than what's on offer via the in camera processing.

You can get significantly sharper edges and corners at the wide end if you are able to elect not to correct the distortion.

Field curvature can play havoc with getting cross frame clarity unless you know where to focus in the frame.

Images greatly benefit from very low radius/high percentage sharpening which brings out the micro contrast.

If you have the accompanying 55 to 210 E mount zoom and you wish to shoot around the 50mm mark go with the 55 to 210 everytime, it is vastly better than the 16-50 at this setting.

Generally you could simply set the aperture to f6.3 right across the entire focal length range and get close to optimal results.





















6 comments:

  1. I would love for Sony to make a similar lens (doubt it would ever be quite as small) for FE cameras but the overly, largely undeserved hate for the 16-50 makes that unlikely. The FE 28-70 is similar in that within a certain shooting envelope it does really well. Thanks for taking the time to so thoroughly test this and put together such a great guide!

    ReplyDelete
  2. Thank you for that, I agree Joel I do think there would be a market for a nice compact pancake for the A7 series, it would be a bit bigger of course of have some compromises, but the practicality would be a neat aspect. On the other hand I guess if you were shooting the A7r mk 2 using the pancake in APSC mode you could still give a very useful 18 megapixels or so and if you turned off the APSC auto cropping it could be more at some focal lengths so as a stop gap for some shooters that might be OK.

    ReplyDelete
  3. Much better is the option, to sell the PZ 16-50 mm if you get one as a kit-lens.

    ReplyDelete
  4. Nice illustrative post, but things could be summarized much more compact.

    For optimal usage of the 1650:
    1. Attach a lens hood to this lens to improve contrast
    2. Use it between 16-35mm, stopped down once from wide open
    3. Turn down NR to "low" incamera.
    4. use Capture One and RAW to get a maximum of 14.9 mm wide-angle lens-corrected

    P.S. Info on lens hood solution for 1650:
    http://luxorphotoart.blogspot.de/2014/05/tipp-diy-lens-hood-for-sel1650pz.html

    ReplyDelete
    Replies
    1. How to attach a lens hood?
      No way I know of

      Delete
    2. Sure you can, you can get a 40.5 mm wide angle hood, metal or plastic, or use a 40.5mm to say 49mm step up ring and attach pretty much any screw in hood. Personally I would go the later and get a collapsible rubber hood. Check out ebay, Crix in the post above has also provided a link to an option. Hope that helps.

      Delete