What a disappointing situation with the 17mm lens! Why?

Started Sep 2, 2013 | Discussions thread
Paul De Bra
Paul De Bra Forum Pro • Posts: 12,473
OK, bar user experience, look just at numbers.

The 25mm is one of the most highly regarded lenses. The 20mm is too but its focusing mechanism isn't state of the art. The 17mm f/1.8 is "not that good"? Let's look at the numbers again:

Lenstip numbers are (read from the graph): 17mm: 65/53, 20mm: 74/58, 25mm: 75/68. What this says is that the 20mm is not far off in the best in the center, worse in the corners. The 17mm is much worse according to Lenstip.

Photozone only has comparable numbers for the 17mm and 25mm. The 17mm is 2832/2134 and the 25mm is 3075/2213. That is a very small difference. Yes the 25mm (and by extrapolation the 20mm) is *a bit* better but not *quite a bit* better.

This is just numbers and while generally I like numbers I also realize that these tests are based on very small samples and may thus not be representative for the average performance to be expected. The numbers on Lenstip versus Photozone indicate that relatively speaking Lenstip got a worse copy of the 17mm than Photozone (assuming their 25mm copies are similar).

There is no denying that the 17mm lens is not quite as good optically as the 20mm. People who have both are confirming that the 20mm is sharper than the 17mm. Yet they tend to feel the 17mm is good enough to keep and some even reported selling their 20mm and keeping the 17mm. This would be completely irrational behavior if the 17mm was cr@p and the 20mm excellent.

Relying on the experience of many non-experts isn't all that strange for someone with a PhD. It is commonly called "wisdom of the crowds" and in data-mining it is called "ensemble learning". It is counterintuitive and remarkable how combining the unreliable opinion of a large number of not very good predictors leads to a highly reliable overall predictor. I know there are caveats such as requiring the predictors to be independent (which is why the theory of wisdom of the crowds and ensemble learning does not mean that the outcome of elections is that the best candidate wins). But still, if a large number of people report one thing and you think they are all wrong you should think again. Maybe you are wrong. Maybe in the case of a lens you (and some review sites) are on the bad end of the sample variation Gaussian curve.

It all comes down to "how much better" Olympus should have been able to make the lens and "at which price". We know from the 45mm f/1.8 that good and inexpensive are possible (but with a lot of sample variation) and from the 75mm f/1.8 that good and expensive are certainly possible (with less reported sample variation). As for the 17mm, maybe it's just somewhere in the middle there...

Anyway, I'm not interested in the 17mm f/1.8. The 20mm I have is good enough and for my next lens I'm waiting for the 12-40 f/2.8.

-- hide signature --

Slowly learning to use the Olympus OM-D E-M5.
Public pictures at http://debra.zenfolio.com/.

 Paul De Bra's gear list:Paul De Bra's gear list
Fujifilm FinePix F200EXR Olympus OM-D E-M5 Olympus E-M5 II Panasonic Lumix G 20mm F1.7 ASPH Olympus M.Zuiko Digital 45mm F1.8 +3 more
Post (hide subjects) Posted by
(unknown member)
(unknown member)
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow