Lenses and Optics

Some M-Mount Field Curvatures

Published August 31, 2014

I haven’t posted very much lately. We’ve had some new equipment installed and we’ve been doing a LOT of testing as we develop our new database of lenses on the optical bench. As the database fills out I’ll be posting more than ever, just because a lot of this stuff is just fun. Today’s post is largely for fun, but will have some additional interest for those who shoot Leica or shoot M-mount lenses adapted to other cameras.

One thing that optical bench testing gives us that is hard to find elsewhere is a clear map of field curvature. We had a client interested in determining field curvatures for a several M-mount lenses and thought there would be a few among you who also wanted to see them.

What These Graphs Are

The graphs are pretty simple: the machine finds the best focus point in the center of the lens (“0” on the vertical axis). It then measures 20 other points from one side to the other of the field, finding the best focus and highest MTF at each point. The relative MTF is shown by color (white>red>orange>yellow, etc.). The focus position compared to best center focus is shown on the vertical axis. The horizontal axis shows position from the left side of an APS-C size sensor to the right.

These lenses were all tested at f/4 to level the playing field for the wider aperture lenses. But this means the field curvature wide open would probably be larger than what you see here. They were done at infinity focus, so the field curvature might be a bit different at shorter focal lengths.

One thing these graphs will show (that you probably don’t really want to know) is that a lot of lenses have a very slight bit of tilt to the field. These are all good copies, tested multiple times. The tilt that is noticeable on these graphs isn’t noticeable in real-world photography, at least not without a great degree of pixel peeping.

The other thing that you may not have thought of is that the sagittal and tangential fields often have different field curvature.

Does this have real-world implications? Yes. The lens with wicked field curvature may give amazingly sharp portraits, but not sharp landscapes or architectural shots, for example. I’m sure someone is going to ask something like, “Well, now many feet does a 100 micron focusing distance equal at infinity?” I don’t have the math to answer that question and don’t have time to go look it all up, but if one of you wants to we’d welcome your input.

Some wide-angle M-mount lenses.

First we’ll show 4 wide-angle lenses. You may notice the Leica 18mm is very mildly tilted, although this is not something you’d notice in a photograph. The Voigtlander 21mm is a good example of a lens with quite different sagittal and tangential curvatures.

A few that are not quite that wide.

The Leica 28mm f/2.8 gives us a nice example, at least in the sagittal field, of a lens with double (sometimes called Sombrero) curvature.

And lastly some 35mm lenses

The Voigtlander 35mm f/1.4 (and remember, this is stopped down to f/4) shows some pretty wicked curvature. Because I know some Voigt fanboy is going to tell me his 35mm has no field curvature I’ll go ahead and tell you that I tested 5 copies and they were all identical. The double field curvature seems to be pretty much standard for these M-mount 35mm lenses.

I don’t have any dramatic conclusions to add, other than I think this is a very useful tool. We’ll be presenting field curvature graphs on all of our lens reports going forward. I’ll also apologize in advance to all of you who want to see the curvature of some specific lens or other. We have over 150 more lenses that need to be tested, minimum of 8 copies of each one, and the zooms at 3 different focal lengths minimum. I’m just not in a position to take requests right now. But we’ll be publishing more of them soon.

Roger Cicala and Aaron Closz

Lensrentals.com

August 2014

Author: Roger Cicala

I’m Roger and I am the founder of Lensrentals.com. Hailed as one of the optic nerds here, I enjoy shooting collimated light through 30X microscope objectives in my spare time. When I do take real pictures I like using something different: a Medium format, or Pentax K1, or a Sony RX1R.

Posted in Lenses and Optics

HandeVision IBELUX 40mm f/0.85

Just The Lenses: Canon vs Nikon Zooms at 70mm

Curtis Patterson

Hi Roger, I love the testing you do. I have a copy of the Voigtlander 50mm f1.2 ASPH M-mount lens and I love it to death, but it appears to suffer from strong field curvature (which may be a function of filter stack thickness when adapted, but not sure). Nobody on the internet has tested the MTF of this lens, nor the field curvature yet. It would be extremely interesting if you could test one or a few copies of this lens!! It’s a very interesting lens – I believe is becoming quite popular, it would be fun to see how it compares to other lenses that have undergone your testing.
NateW

Out of curiosity, which lens is the “Leica 35 f/2”? I’m assuming it’s the pre-asph “version iv” because of how Leica touts the flatness of the field in the ASPH version? I have the pre-asph and am just trying to learn as much about it as I can. Thanks!
Scott Kirkpatrick

Also, reading carefully through the comments, I can identify all the lenses tested except the Leica 21/3.5 and the 28/2.8. Was the first in fact the recent 21/3.4, a highly optimized Karbe design, or something older? Was the second an older 28 Elmarit design (there are several), or the 28 Elmarit-asph, introduced about the same time as the M8 first appeared? That’s sort of an “early-Karbe” design, and looking at Leica’s specs you could suspect that it was only designed to do best on the smaller M8 image size. I think the wide straight-lipped field curvature images correspond with the latest designs that had the most computing power thrown at them, while the Joker smiles and sombreros are earlier designs which stuck closer to previous seat-of-the-pants layouts with well-loved “signatures.”
Scott Kirkpatrick

Oops, I asked about M lenses in your Canikon thread, not having seen this article. But it would still be interesting to see the most recent Leica 35/1.4 design, with a floating element and possibly other changes. Was this one measured?
Mr Will

Can I be the Voigt fan-boy? I have one and it’s an awesome lens.

Well; apart from the field curvature (which definitely happens), the focus shift (which definitely happens), the barrel distortion (which definitely happens) and the softness wide open (which, you guessed it, definitely happens).

But apart from all those things it’s an awesome lens… 😉
Ilya Zakharevich

BTW, another nice geometric tidbit: the field curvature IN THE SUBJECT SPACE is the same as in the image space. So just take the curvature of the Roger’s images, and this is the curvature you will get in the subject space!

For example, with my DPI, Roger’s images expand 100?m to 6mm (vertically), and expand 30mm to 36mm (horizontally). The vertical expansion exaggerates the curvature 60 times; the horizontal expansion decreases the curvature 1.2²=1.44 times. The total is about 40x exaggeration of the curvature.

Take the last image; the curvature “of the brightest part” of the image has radius about 2cm (I did not try to be precise in my measurements!); so the real curvature near the sensor plane is 40*2cm = 80cm. So a picture drawn on a beach ball of radius 80cm is going to be in focus (as far as it remains on the part of the sensor where curvature is directed up) no matter on which distance from the lens this ball actually is.

[I assume that UP on the Roger’s images is IN THE DIRECTION TO THE SENSOR. And my argument was in the approximation that the field curvature NEAR THE SENSOR does not depend on the focal distance.]
Ilya Zakharevich

At finite distance, the exact formula (for small values of 100?m 😉 is
Focus shift = 100?m / Magnification²
For example, with 50mm lens at distance 5m, magnification is 1/100, and focus shift is 1m. For infinity focus, the value 100?m is always too large; so one may want to find distance D on which the formula above gives focus shift of D.

Example: at 25m, magnification is 1/500, and the formula for the focus shift gives 25m; doing 10 small increments of 10?m starting at this distance would lead you close to infinity. (Exact process could involve integration instead of taking many small steps.)

As you can see, this matches what Jose said!

(For nitpickers, and short focus distances: the formula IS exact, but one may need to be more careful finding the magnification. With the example above, the actual magnification will be not 50mm/5m, but ?50.5mm/5m, since the image will be formed about 0.5mm behind the focal plane.)
David

Voigt fan boys will say what they will. To me it just looks like the happiest 35mm lens in the bunch. Also I wonder why the ZM is so sad 🙂
Capablanca

The calculations ware done for 100 microns. The “10 microns” in the text is a typo, should be “100 microns”.
Capablanca

“how many feet does a 100 micron focusing distance equal at infinity?”
The answer is: Infinity.
That’s because Infinity minus a finite distance = Infinity.

At finite distances, one can compute as follows.
Say 2 meters away from the an 18 mm lens. The lens-sensor distance will be
1/(1/18-1/2000) = 18.163471 mm. If the lens is moved 10 microns (0.1 mm) away from the sensor, the focus point will be
1/(1/8-1/(18.163471+0.1)) = 1248 mm = 1.25 meters away from the lens.
If the lens is moved 10 micron (0.1 mm) toward the sensor, the focus point will be
1/(1/8-1/(18.163471-0.1)) = 5.12 meters away from the lens.

So the equivalent distance is 75 cm towards the lens and 312 cm toward Infinity.

However, at finite distances, the amount of shift in “focus position” (the vertical axis in the plots) may be different from that at infinity, and if that is the case, these computations won’t be relevant.
Samuel H

It’s a pity you didn’t test it, but I have the impression that my Leitz Summicron-C 40mm f/2.0 for Leica-M mount has some pretty wicked field curvature (sides focused a lot farther than center).
Mike

I like the charts . . . but unfortunately (since I don’t really know what to look for) they don’t mean a lot to me.
Interesting, nonetheless . . .
Oskar Ojala

Thanks! This is quite useful, since testing for field curvature is tedious without a bench setup and the results here can be useful in udnerstanding different behaviors.

Would be interested to see if SLR lenses have similar trends or are they flatter or less flat.
grubernd

and that double-M is kind of the signature look of those lenses shot wide open. the whole field curvature thing will also show which lenses are good for focus-recompose, which are not and what lens to use for flat reproductions, including brickwalls and other standard test targets.

meaning: unless you test a lens against a 3-dimensional target specifically designed for that particular lens you will never be able to evaluate it’s “corner sharpness” or whatever in any meaningful way.

oh.. you could use an optical bench or something like that and print pretty colorful graphs, that would work.

thanks, thanks, thanks.
Donald E. Dunbar Jr.

The graphs provide a bunch of data. It seems to me that it would be of assistance in evaluating the graphs if you would publish a graph of a theoretically perfect lens so it could be compared against the actual graphs. Thanks for all the great work.
Ron

Thanks Roger. Does it mean these graphs are with no glass in the optical path?

I think many of these lenses would be pre-digital designs, except the Voigtlander 21/1.8, Leica 21/3.4, 24/3.8 and probably also the somewhat older Leica 28/2.8 and 35/2.5 (which is what I assume is labeled as Leica 35/2.8?). It would be interesting if the newer lenses were better without glass, since I’d expect these will mostly be used on digital going forward. It would also be interesting to see the measurements for the different glass thickness. Perhaps a consideration for future posts?

I’m not surprised with the CV35/1.4 performance. Its twin, the 40/1.4 also exhibits a huge degree of field curvature.

One more question:

Other than the first three graphs, it looks like the remainder were measured to 20mm image width, which would then be more or less full frame?
Roger Cicala

Ron, I apologize, my labels were pretty cryptic – the original work was done for another company and they had provided the list of lenses they wanted tested so I didn’t have to identify thoroughly for them. They were the Zeiss 21mm f/2.8 Biogon, the Voigtlander21mm f/1.8 Ultron.

I’ll also add that we ran these lenses at 0, 1, and 2mm of glass in the optical path. Not surprisingly, since these are rangefinder lenses, we found they were all best with no glass.
Ron

This is extremely informative and is great to see because it very much confirms my experience with some of these lenses. Thanks for starting this off with rangefinder lenses!

A couple questions:

Which Voigtlander and Zeiss 21mm rangefinder lenses are these? Each brand makes two models.
Are the measurements made with just the lens, or is a thin piece of glass also used to simulate sensor stack thickness? If the latter, what thickness?

In the Zeiss paper CLB 41 “From the series of articles on lens names: Distagon, Biogon and Hologon” by H. H. Nasse, he states:

“If the filter is significantly thicker, the contrast transfer for the image edge becomes worse for tangential structures. In the graph of the curves, this looks like the old retrofocus lenses but is caused by astigmatism rather than lateral chromatic aberration. The focus is shifted to greater distances for tangential structures by the additional path through the glass.”

By filter, he’s referring to the glass of the sensor stack.

Therefore, with thicker sensor stacks it won’t be unusual to see significantly different tangential performance.
Tim Ashley

Utterly awesome. This will really help visualise something that one has to try and learn slowly, per lens. It’s such a slippery process and your new graphs will provide great traction. Thank you Roger!
Jose

“I’m sure someone is going to ask something like, “Well, now many feet does a 100 micron focusing distance equal at infinity?” I don’t have the math to answer that question and don’t have time to go look it all up, but if one of you wants to we’d welcome your input.”

I think the answer is:

Distance(mm) = 10x(FL)^2, where FL = focal length in millimeters
Louis

You guys are awesome. Whether or not there are real photographic implications to some of the variations is irrelevant…. the data itself is just fascinating.