Friday, 20 June 2014

Maverick lenses - the Panaleica Summilux 15/1.7 for m4/3

It is an Urban Legend among m4/3 users that no matter what, Panasonic and Olympus don't succeed in making a sharp lens around the the 17mm (35 mm eq.) focal.
It seems that the legend is now hitting again the prestigious Panaleica 15/1.7.

DPR readers complained that the Oly 17/1.7 was not sharp enough, and now a comparison at DxOmark is showing that the Panaleica, while a bit better, is not by much. 
Note that the lens is not exactly cheap: some 600 USD, and 620 Pounds in the UK! It does have a very pretty bokeh, a nice colour and tone signature, it is resistant to flare, but  the Sharpness or Death Gang are deaf to these niceties. OTH commercial sites are waxing lyrical:

Photography Blog review

But, well, they are commercial sites.

Note that at the edges of the 17mm, the Panny 14mm/2.5 and the 20mm/1.7 have no such problems, they are considered very sharp by the same crowd.
IMHO user samples speak for themselves. From a DPR user, Zilver, the 15mm at 1.7, and 4.0 respectively (open the piccie in a new page to get the larger shot):
Check the foliage in the foreground and compare with the buildings in the backround, especially at the edges. There is a considerable difference between the two shots. 

The problem seems to occur at infinity, less so at short distance. See this pipe pictures at DPR by Eastvillager:

here f/1.7 is OK. Curvature is less important in the center, and at close range.
So what happened? Both in the case of the Panaleica 15/1.7, and in the case of the Oly 17/1.7 curvature problems have been mentioned by the reviewers. That points a finger to firmware correction being insufficient.
Let me explain the general case. Lenses are curved, and thus give a curved image, but their projection on the sensor must be rectilinear, as in a Mercator Map, reflecting a Globe.

You have two ways of doing that: to associate a concave lens to the convex main one - to simplify things.  Or to use firmware correction, like you do when you want to straighten a fisheye. That is exactly what happens in my 14/2.5. 
I remember a reviewer showed that the FOV before correction is 11mm (in RAW) and narrows to 14mm after correction. 
That means that edges have been cut away by the manufacturer because they were fuzzy.  See the m43photo review of the 14mm.

In m4/3 the problem is compounded by the short Distance to Flange, which gives less elbow room for correction, the geometrical problem being more severe.
Finally this gives me a cue about what might have happened. With its high optical standards Leica might have imposed to Panasonic to use as little firmware correction it could.

As a result, to keep the lens small Panny left some geometrical distortion uncorrected. I say this because the same has been argued about the Oly 17/1.8.  If you look at the edges, it's not that the pixels are distorted, but they are out of focus! That is to say that because of geometrical distortion in a WA you cannot have both the center, and the edges in focus!

Now the ideal of the old 4/3 was always to be 'even across the frame', but the distance to flange was double what it is in m4/3. More restrictively, they never attempted to go below f/2.0.  With a wider aperture,  focus problems increase.

The Panny GX7 with the new Panaleica 15/1.7. Yummy!

To get back to the Panaleica 15/1.7 there is no sharpness problem at f/4, but there is one at 1.7, at the edges. Close range is OK.  So what does it tell about the use of the lens? 

The 30mm eq. should be fine for Street Shooting. It should be fine for low light Interior Portrait shooting. Its good bokeh should be nice for the occasional flower, but it's not so good for Landscape, unless you stop it down.

This in my view is a minor letdown, but certainly I won't sell my ultrasharp 14/2.5 (see shot in the post above), nor my Sigma 19mm/2.8 which is quite sharp full open. 
I bought both for a song, I am v. happy with them.
If I didn't have them, then perhaps I would consider the Panaleica 15/1.7, but mainly for its lovely colour and tone signature. My lenses are less filmlike in their results, more digital, although I can use Olympus art filters to overcome the coldness of digital rendering.

Sharpness for me was never a fixation. Good to have, but not decisive. As HCB famously said:
"I'm always amused by the idea that certain people have about technique, which translate into an immoderate taste for the sharpness of the image. It is a passion for detail, for perfection, or do they hope to get closer to reality with this trompe I’oeil? They are, by the way, as far away from the real issues as other generations of photographers were when they obscured their subject in soft-focus effects." -- Henri Cartier-Bresson

Reviewers say that the lens  is well built in light metal, and that the aperture ring is a godsend, but only on Panny cameras. With Olympus it doesn't work. m4/3 is a Standard of sorts - oh well...

Post Scriptum

 Jordan Steele whose opinion I value has a different explanation, here, after looking at  the differences between RAW corrected and uncorrected images, which I didn't do. 

 He argues that indeed the lens has a lot of curvature and that the fuzzy edges are due to over-correction by FW. The conclusions are much the same, although the reason is different. Panasonic dealt with curvature, the same it did with the (now) much cheaper 14/2.5 pancake.

I still think that Panny should have ignored bokeh requests, and make instead an f/2.8 outstanding lens.


  1. The aperture dial not working on Olympus cameras is quite a let down, I really hope Olympus decide to accommodate this in future cameras, or the standard is somehow redefined to incorporate aperture dials. It's one of the things I admire in the Fuji X system, and one of the reason I love using MF lenses on M43. Nice article, I enjoyed the thoughtful review.

  2. This is an era of declining sales of bodies, so mftrs, are desperate to sell their own lenses. Unfortunately what suits them, doesn't suit us. Never say never - this is a case where a firmware decision can fix the problem. As you mention, in the long run it pays.

  3. Very good review. Loved the discussion of the effects of curvature. Well done.

  4. Thank you. In fact there are 15mm fisheyes, and when you try to straighten them up, you get bad edges that you must cut away. That was my lead