Noctilux/Summilux/Summicron Comparison

Several weeks ago, I posted the results of a comparison I had done between a Noctilux and a Summicron, and mentioned I wished I had had a Summilux to test at the same time… Last week I came across an opportunity on a Summilux and took it. So, for those who are interested, I re-shot the test targets, this time with all three lenses, and offer the results below. The lenses I compared were a fifth version Summicron, which is the last tabbed model with the removable/reversible hood; a current version Summilux, with the built-in lens hood; and a second version Noctilux (Canadian), with the removable hood. The same M6 TTL body was used for all shots, the camera was mounted on a tripod, and the film used was Velvia. Note: The new results for the Noctilux and the Summicron did not vary significantly from the prior comparison, but having the Summilux in the mix required me paying more attention to the subtleties of the differences. Again, these are nothing more than my impressions, and I share them for those who have an interest in this type of discussion.

At f1, I rate the sharpness of the Noctilux as "Good" at the center and edges. There is noticeable vignetting of about 2 stops at the extreme corners at f1.

At 1.4, the Noctilux center sharpness moves up to "Very Good", while the edges remain "Good". Interestingly, the performance of the Summilux is essentially the same as the Noctilux at this aperture. The vignetting on the Noctilux has been greatly reduced, but is still visible, and is a bit more than that on the Summilux.

At f2, the Noctilux and the Summilux remain equal in the center at "VG", and are almost as good as the Summicron which I rate as "VG+". The Summicron is "VG" at the edges, and better than the Summilux which has moved up to "G+" and the Noctilux which remains a "G". There is a slight, and virtually identical amount of vignetting visible in all three lenses.

At f2.8, the Noctilux now gets a "VG" for center and edge sharpness. The Summilux and Summicron are essentially equal in the center and have moved up to an "Excellent". The Summilux now gets a "VG" at the edge, but the Summicron has improved as well, and getting an "E", and the Summicron now exhibits its trademark of uniform sharpness corner to corner. There is a trace of vignetting visible in the extreme corners of the Noctilux and the Summicron, but interestingly none was detected in the Summilux.

At f4 the lenses continue their performance and all get the same ratings as for f2.8 above. No vignetting was noticed in any of the designs.

At f5.6, the Summicron and the Summilux are still "E" in the center, and the Summilux has gained a bit at the edges and is almost, but still not quite equal to the Summicron. The Noctilux has lost a little ground, but is still "VG" at the center and edges.

At f8, the smaller aperture has begun to take its toll on the designs, but for some reason to a slightly lesser degree on the Noctilux, so that in essence, all three lenses now rate a "VG" at the center and the edge.

At f11, and f16, the smaller apertures continue taking their toll on all three designs, and for all practical purposes, the three lenses perform at essentially the same level, rating "VG’s" at the center and edges.

Both the Summicron and Summilux are Leica-neutral in colorcast. The Noctilux, on the other hand, runs slightly to the yellow-warm. I learned recently that Leica uses a special UV glass in the Noctilux because its refractive index was ideal for the lens computation, and could also explain the warmer cast.

While bokeh on the Summicron is very smooth and pleasing, I found it even (slightly) more so on the Summilux. On the Noctilux, the bokeh has an almost "liquid" smoothness to it that I think imparts that special "look" the Noctilux is known for.

So, which one is the best? Tough question. All of these lenses perform at a very high level, and all have their specific strengths. The Summicron is the sharpest corner-to-corner at most working apertures, the Noctilux excels in low light (and high contrast light), and the Summilux is somewhat of a well-sized compromise between the two. For my style of shooting I could probably make a good case for choosing any one of the three. In fact, based on the results of my earlier test comparing the Noctilux and the Summicron, I made a case for keeping both. However, when push came to shove, I almost always grabbed the Summicron over the Noctilux to round out my shooting bag because of its compact size. The only problem was, more than once I found myself looking for an extra stop on the Summicron that wasn’t there. So, I’ve decided to go with the compromise. I’m giving up the corner-to-corner sharpness of the Summicron and the extra stop and "look" of the Noctilux in the interests of having maximum versatility and keeping the weight and complexity - not to mention cost - of my M system to a minimum by choosing one lens –- the Summilux.

-- Jack Flesher (jbflesher@msn.com), April 28, 2001

Answers

Thanks, Jack. You used the current Summilux. I wonder how much of your observation can be attributed to recent glass, coatings, etc, versus optical design (which, I assume, should be characteristic of other recent, but not current, versions of the lens)?

I have the #11114 lens (c.1980) with detatchable hood. Any "gut feeling"? I suppose that would be just a wild guess/assumption.

-- Ken Shipman (kennyshipman@aol.com), April 28, 2001.

The Noctilux has a focus shift problem at apertures smaller than f1 that makes it, in my opinion, unacceptable. I've owned one, and found the focus shift to spoil portraits at, say f2. At f1 it really is incredibly useful, and mine focussed perfectly at full aperture, on two different m6 bodies. See Erwin's Photosite at

http://www.imx.nl/photosite/leica/mseries/testm/M10-50.html

for details of this phenomenon. He considers that the additional depth of field as the aperture is stopped down generally compensates for the focus shift, but it didn't for me. It probably depends on the subject distance- I was photographing people at 2.5 metres. I borrowed a different Noctilux and it was the same. If you want to see some interesting test negs, email me.

-- john Stockdale (jjss@bigpond.net.au), April 28, 2001.

Now We have a round answer, I missed the ´lux in your last comparison, now that I read it, your conclusions seems logical, just a question, how large were the prints or proyections from wich you made this comparisons?, I would like to know how you did it.Any way thanks for sharing.R watson

-- R Watson (al1231234@hotmail.com), April 28, 2001.

Ken: As I understand it, Leica has revised the optical formula of only the Summilux in the latest version of that lens, and that the optical formulas in the versions of the Summicron and Noctilux I own are identical to the latest offerings.

John: Yes, I've read Erwin's article, and am aware of the focus-shift problem at smaller apertures with the Noctilux. However, it is only a very slight shift, which can affect your results when doing flat copy work and lens tests to be sure. But as a practical matter when shooting three-dimensional objects, it is not a critical issue as the plane of precise focus exists somewhere very close to where you originally focussed in the 3D image. We are talking microns of focus shift, which translate to a few milimeters of plane-of-focus shift in the 3d image.

R. Watson: I did not make prints. I made the comparisons by directly viewing first-generation transparancies with a 16x *Peak* loupe. (Velvia is slide film.) With a print, I'd be viewing a second- generation product which can introduce the possibility of 3rd party error from either the optical system of the enlarger, or the person doing the printing.

-- Jack Flesher (jbflesher@msn.com), April 28, 2001.

Jack; it seems so small to be so critical,IMOO, 16X loupe is powerful but...well any way I don´t think you´re far from real,so what do you think?, I don´t want to be critical at all, apreciate your test and your time, but since I love big prints, it is always my way of apreciation, I own a summicron last gen. and when bougth it I was so tempting to buy the ´lux instead, I´m very satisficed with the ´cron, but for a 1.4, I think my option will be the 35 asph. and keep the ´cron for prime quality, well time and money will tell.I liked your direct comparisons side by side, E. Puts´is very critical, but is hard to imagine it side by side, I always feel a diferent criteria for each lens.Thank´s Jack

-- R Watson (al1231234@hotmail.com), April 28, 2001.

The thing about this test, just like Erwin's tests, is that the test may be done with scientific method, but the analyses of the results are subjective. There are no reference points from which the reader can correlate to his own standards. As sterile as MTF graphs and lppm resolution figures are, and even if they don't say much about real-world photography by themselves, they do give a quantitative indication of how one lens compares to another. In the case of the 50/2 vs 50/1.4, the published results at www.photodo.com bear out pretty much what Jack says, and what my own opinion happens to be, so I would feel confident that Jack and I would probably be on the same page regarding other lenses as well. But without the numerical data as a standard of reference, it all comes down to trusting someone else's eyes and standards, which I think is a kind of risky way to research a lens purchase.

-- Jay (infinitydt@aol.com), April 28, 2001.

IMO, 16x loup is not enough to check out lens performance

a 20/20 eye can resolves about 4 -5 lpmm, with a 16 loupe, the the maximum resolution is about 80 lpmm, which is not a high number

I use an made in Wetlzlar Octscope with eight different magnifications, which goes to maximum of 28x

For check out lpmm of lens with test target, I use an Olympus microscope at 100x.

-- martin tai (martin.tai@capcanada.com), April 28, 2001.

Jack Flasher:

Thank you for your excellent contribution. I think that you are correct, in the final analysis, each M-shooter has to make their own choice balancing all those things you dicussed, plus price and the other lenses they will use in conjuntion with the 50mm. Very tough to do, and always fun to do too.

-- Dan Brown (brpatent@swbell.net), April 28, 2001.

My immediate apology to Jack Flesher for misspelling his name.

-- Dan Brown (brpatent@swbell.net), April 28, 2001.

Jack, with respect, I'd like to challenge your statement about the Noctilux focus shift, where you say "We are talking microns of focus shift, which translate to a few milimeters of plane-of-focus shift in the 3d image." Erwin gives figures for the focus shift of 74micron at f2 and 120micron at f5.6, but seems uncharacteristically vague when he describes the effect of this degree of focus shift. After all, a micron seems pretty small! (One thousandth of a millimeter).

My experience is that it really makes a difference at around f2 to f5.6. My use of the Noctilux was to use it in the range f1 to f2.8. The ability to get more than f1.4 in available darkness situations was a real luxury. It is a tribute to Leica that the M6 focusses the Noctilux practically perfectly at f1. But where possible, I'd like to use a smaller aperture than f1 if I could: that was my problem.

At the following focussed distances the actual distance focussed, using Erwin's figures above, are as follows:

At f2: 2 metre: 2.12, discrepancy = 12cm = 4.7inch 3 metre: 3.28, discrepancy = 28cm = 11inch 4 metre: 4.52, discrepancy = 52cm = 20inch 8 metre: 10.45, discrepancy = 2.45metre = 8feet

This was calculated using the formula we know and love, the lens equation: 1/u + 1/v = 1/f It seems amazing doesn't it, but when you think about it, a lens doesn't move much when you change focussed distance from 3metre to 3.28metre, does it! Now although these figures seem amazing, you can verify them by photographing a 3-dimensional target with your Noctilux. Photograph a scale with it and measure it.

Now, we need some depth of field for the following reasons: 1. Our subjects are 3-dimensional 2. Our cameras are not always perfect 3. Even the photographer is not perfect at judging the exact focus. 4. Our subjects are not immobile.

If you view the above focus shifts to be accomodated satisfactorily by depth of field, you are: 1. Accepting as sharp a circle of confusion of over 35 micron, which was recommended in the 1930s when films and lenses were very inferior to today. (I know this opens up room for lots of debate, but we didn't buy Leica gear to get that level of fuzziness) 2. Even if that level of fuzziness is ok with you, you have zero margin of error on the near side of the focussed distance. (And heaps on the other side) This was unsatisfactory for me.

-- john stockdale (jjss@bigpond.net.au), April 29, 2001.

The table in the message above should have looked like this:

At f2:

2 metre: 2.12, discrepancy = 12cm = 4.7inch

3 metre: 3.28, discrepancy = 28cm = 11inch

4 metre: 4.52, discrepancy = 52cm = 20inch

8 metre: 10.45, discrepancy = 2.45metre = 8feet

I didn't realise that the line feeds would disappear.

-- john stockdale (jjss@bigpond.net.au), April 29, 2001.