In an earlier post we were discussing metering sensitivity at low levels of light. I had read a stretch of Gunter Osterloh's book on the M that discussed this in a somewhat confusing manner. Well I have read and re-read the whole section now a few times, and here is what I take away from it:

Apparently, since all lenses vignette to some degree at any aperture, and since vignetting is most prevalent in fast lenses at wide apertures, you don't get the whole enchalada when it comes to fast f-stops as the vignetting cuts down on the lens' ability to gather light across the entire image! Mr. Osterloh uses the example of the Noctilux, and claims that while the maximum geometric (measured) aperture is indeed f1.0, the maximum sensiometric (actual light transmission) aperture is something closer to f1.2 or 1.3 (although he does not say it that specifically, which is why the confusion), HENCE YOU WILL OFTEN NOT SEE A METERING DIFFERENCE WHEN YOU CHANGE FROM f1.4 TO f1.2 OR f1.2 TO f1.0!!! And it gets better -- but the DOF will be equivalent to the geometric aperture, so in effect you lose DOF faster than you gain light!

Has anybody heard anything like this before? I am having difficulty believing that when you pay for a Noctilux, you are not getting everything you thought you were paying for. If it is true, then I think Leica has committed the greatest marketing coup of the century!

-- Jack Flesher (jbflesher@msn.com), December 23, 2001

Answers

I think the key is your statement, "to gather light across the entire image." Vignetting affects the field of the image primarily. Under certain conditions it can also affect the metering area. So what happens is that while the meter indicates f1.2 the center of the metering area is actually f/1. If you expose at f/1 the very center of you neg will have the correct density but it fall off sharply as you get into the field. If you expose at the metered f/1.2, then the very center of the neg will be overexposed slightly with the field having more density as well. A Leica-User, Henning Wulff, went through all this with extensive tests. The Noctilux is a true f/1 lens (a statement Erwin backs up as well) but a certain amount of vignetting was left in the design to control aberrations from skew rays. This vignetting, under certain lighting conditions causes the meter to behave as you have noted.

Great explanation but what about the real world. I would tend to meter and go. A little extra density in the center is not going to hurt.

-- John Collier (jbcollier@powersurfr.com), December 23, 2001.

Jack: I have not heard of this, nor do I own a Noctilux lens. However, for those that do, a simple test would be to remove the lens, and meter on an 18% gray (or any uniform object) and determine what shutter speed is required, then try the same thing with a Noctilux in place.

Looking at your statement as to losing DOF faster than light, I would believe this after thinking about it as the greatest curvature is near the edge. Looking up a 50mm lens in the tables in Andrew Matheson's "The Leica Leicaflex Way, Tenth Edition", the following shows up: 50 mm lens set at 10 feet. f:1.2 DOF from 9'7" to 10'6", 11" f:1.4 DOF from 9'5" to 10'8", 1'3" yields a 36% gain over f:1.2 f:2.0 DOF from 9'3" to 11'0", 1'9" =40% gain over f:1.4 f:2.8 DOF from 9'0" to 11'4", 2'4" =33% gain over f:2.0 f:4.0 DOF from 8'8" to 12'0", 3'4" =43% gain over f:2.8 f:5.6 DOF from 8'2" to 12'10", 4'8" =40% gain over f:4.0 f:8.0 DOF from 7'6" to 15'0", 7'6" =61% gain over f:5.6 and this does nothing except add confusion to the above, but I will leave it in.

H..E...L....P.....?

-- Mark A. Johnson (logic@gci.net), December 23, 2001.

Yes Jack. You read it correctly, vignetting (natural) is present in the Noctilux at f/1.0: this is unavoidable, the laws of physics applying equally in Canada, Germany, and Japan. However, you will see if you read Pop Photos camera/lens tests that at low light levels with a lens wide open you always lose 1/3-1/2 stop because of vignetting, so an f/1.0 lens is still about 1 stop faster than an f/1.4 lens.

In addition, Osterloh makes the point that most large aperture lenses also deviate from their geometric f stops because of loss of light transmission due to ineffective coatings. Leica lenses show less preventable loss of light transmission than competitors lenses tested at the factory. Thus, for two lenses of the same nominal aperture (eg., f/1.4), one may transmit 1/2 stop or more than the other.

The answer is you get what you pay for when you buy a Noctilux: a geometric aperture of f/1.0 with artificial (avoidable) vignetting reduced to an absolute minimum, but natural vignetting still present. Also, because the distance travelled by light impinging on the corners of the film plane is greater than that impinging on the center, the very center of the field may be properly exposed while the corners may be slightly darker at f/1.0.

Finally, if you look at Pop Photo's lens tests, in some, the measured geometric aperture is less than the stated geometric aperture for Japanese lenses. I doubt this is the case for the Noctilux. Anyway, try and find another f/1.0 lens that transmits more light wide open than the Noctilux. Good lucl!

-- Eliot (erosen@lij.edu), December 23, 2001.

Jesus! This is why some like me welcomed the light hearted relief of Alfie! I mean:

......10 feet. f:1.2 DOF from 9'7" to 10'6", 11" f:1.4 DOF from 9'5" to 10'8", 1'3" yields a 36% gain over f:1.2 f:2.0 DOF from 9'3" to 11'0", 1'9" =40% gain over f:1.4 f:2.8 DOF from 9'0" to 11'4", 2'4" =33% gain over f:2.0 f:4.0 DOF from 8'8" to 12'0", 3'4" =43% gain over f:2.8 f:5.6 DOF from 8'2" to 12'10", 4'8" =40% gain over f:4.0 f:8.0 DOF from 7'6" to 15'0", 7'6" =61% gain.....

I mean, come on - who cares!

-- Giles Poilu (giles@monpoilu.icom43.net), December 23, 2001.

Sorry Giles - gotta take the bait ;-) This is a far more reasonable question than "isn't this poorly composed, terribly lit, uninteresting snap of the princess a great photo because it touches the angst of the primordial slew....yada yada yada?"

-- Bob Todrick (bobtodrick@yahoo.com), December 23, 2001.

I refuse to re-take the bait, I am observing our Christmas truce!

Happy Christmas Bob!

-- Giles Poilu (giles@monpoilu.icom43.net), December 23, 2001.

John and Eliot: Okay... If I am to understand you, Erwin and Osterloh, then all of this is a perfectly normal optical phenomenon. So a Noctilux is really f1.0 but transmits more like f1.3, and it is all okay because the Summilux really doesn't transmit f1.4 either, but something more like f1.6. So what does the Summicron actually transmit, f2.2?

Mark: Fair question (or confusion!) and I do think it gets very confusing... All I was trying to say here is that the geometric aperture is used for the DOF calculation but the sensiometric aperture is what reaches the film, so in the case of fast lenses you have the limited DOF of an actual aperture that is not really transmitting its fair share of the light.

Giles: I thought we were beyond all of this. Quite frankly, you are now the one that is very OT. I think it is becoming patently clear that you take some perverse pleasure in stirring up shit. It appears you really miss Alfie not for his inputs, but for the grist he provided for your own personal mill. Perhaps it is you we should be pleading with Tony to "moderate"... Or then again, maybe you just need to get laid.

Bob: Well said... I agree that Mark did poses a fair question.

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Giles: After re-reading my above response to you, I appologize for that last comment -- it is not my usual style and two rudes do not make a polite.

Peace and Happy Holidays,

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Hmmm...please Jack, that is absolutely not my intention. If you read my posts across various topics together with my reply to Bob above you will hopefully see this. Don't let it be said however that I don't enjoy a good argument!

I also say that I enjoy technical meanderings as much as the next man, but this thread is just a little too much, even for me.

Lastly, thank you for the offer but my wife kindly took care of this not 10 hours ago.

-- Giles Poilu (giles@monpoilu.icom43.net), December 23, 2001.

Giles & Jack

I input the data and hoped it would be in a tabular format, and when it wasn't it made for a bigger mess than Jack's original question. Thanks all for the input.

Merry Christmas to all. :)

-- Mark A. Johnson (logic@gci.net), December 23, 2001.

Jack

You got it. The marked (geometric) aperture is just a nominal figure, and not all of these marked apertures are even correct as marked. Leica has always maintained that their "effective" apertures are higher than those of the competition for their fast lenses. I believe this is probably the case (or at least it was the case). Anyway, if you use the Noctilux regularly, you can't help being impressed with the light gathering power of this lens. Its low light capabilities are actually greater than the vast majority of us will ever need. Leica says it will reproduce details not even visible to the photographer at the time the photograph is taken! Don't doubt it.

-- Eliot (erosen@lij.edu), December 23, 2001.

Giles: above all I agree - Happy Christmas to you and your family. And if your wife did take care of 'that problem' - what more can anyone wish for. MERRY CHRISTMAS TO ALL AND THEIR FAMILIES - I think I'm actually going to sit and watch "It's a Wonderful Life" tonight. Hope there's lots of Leica gear under the tree this year.

-- Bob Todrick (bobtodrick@yahoo.com), December 23, 2001.

The sainted Erwin sez the corner fall-off with the Noct is THREE stops.

If you look at a Noct f/1 shot with a background that is evenly OOF, but has "things" in it, you can see that the circles of confusion get visibly smaller near the corners. The depth of field near the corners is actually that of an f/2.8 lens, or thereabouts.

It's where those "swirly" backgrounds come from that we were commenting on a couple of months ago.

Shoot something with a constant width OOF (like, say, a long flourescent tube (or bank of tubes) across the top of the frame, and the blur at the ends will be substantially more pinched than near the center - what should be a soft rectangle is actually projected as a soft diamond shape.

Sometimes it's useful to take an unloaded camera and open the shutter on "B" and look through the lens. With the Noct (and most other fast lenses), by looking through the corner of the film gate you can see just how much of the circular opening gets "cropped" into a "cat's-eye" shape by the barrel and/or the optics themselves.

You can also stop the lens down and see for yourself at which aperture the whole pupil starts to become visible, which is where vignetting will disappear**. This can be especially useful in determining whether, e.g., a teleconverter will cause corner darkening with a given primary lens.

A lot of Mandler's lens designs have a fair amount of light fall-off. Erwin also sez this was an intentional trade-off to maximize center sharpness and contrast with the glasses and other technology he had available 20-40 years ago. FWIW.

**(except for the square-root fall-off from wide-angles like the Super- Angulon, as Eliot points out.)

-- Andy Piper (apidens@denver.infi.net), December 23, 2001.

Eliot: I agree that the Noctilux is indeed a light-gathering machine. However, I noticed with mine (I've owned and sold two) that I could use the same shutter speed while shooting at f1.4 and at f1 (when the meter agreed with both) on slide film, and get transparencies that showed very little difference in density. They showed a difference, but it was nowhere near a full stop, but more like 1/2 stop... Hence my overall confusion on this issue.

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Jack, I'm going to confuse the issue even more: If a full-screen averaging meter suggests the same shutter speed at maximum aperture as at one stop down due to vignetting at the outer zones at maximum aperture, this is understandable. However the M6 reads only from the white spot on the shutter curtain which is in the center of the frame, so shouldn't it be unaffected by light falloff outside that area? And, how is it that on an SLR where the meter always reads the light at maximum aperture, the meter is "told" the shooting aperture via cams and has no clue about light falloff, the meter indicates a full shutter speed between each stop and we still get proper exposure at all apertures?

-- Jay (infinitydt@aol.com), December 23, 2001.

Jack

It is NOT my experience that the meter indicates the same shutter speed at f/1.0 and f/1.4. I have just not seen this. If it did happen, the f/1.0 shot should be better exposed. You have two lenses (50/1.0 Nocti and 50/1.4 Summilux) with the same focal length and lensmount, for the same camera, one of which has a lot more glass and a much larger entrance pupil (Noctilux). It does NOT make sense that these two lenses transmit the same amount of light to the film plane at maximum aperture. It's counter-intuitive.

As far as Jay's question, the coupling cam tells the camera the maximum aperature. I wouldn't assume that the number transmitted for an f/1.4 is f/1.4. It might be higher to compensate for light falloff. Alternatively, SLRs may underexpose at maximum aperture. If you read Popular Photography's camera reviews, when they review the lightmeter accuracy, inevitably the report says underexposure by 1/3 up to 3/5 of a stop.

Unless you do careful testing of the lightmeter with large aperature lenses like the Noctilux, I cannot accept off the cuff remarks about someone remebering this or that from several years ago. If you don't like the Noctilux, fine. It's not everyone's cup of tea. But it is the best lens of its type, and I do not find in any way that Leica's advertising is misleading.

-- Eliot (erosen@lij.edu), December 23, 2001.

Jay:

As to your first question, keep in mind the entire lens traslates a single point in a scene to a single point on the film via a light ray, and the scene is composed of many such rays all using the full aperture of the lens. In this sense, the aperture is just a light valve, so the number of points that are transmitted to the film do not change, only their "power" as the ray is squeezed by the aperture. Hence, the f-stop selected has no bearing on where the meter takes its reading, and any vignetting present will affect every light ray that falls on the film, even those at the center.

As to your second question, Osterloh spoke to this in his book in a special footnote. He reiterated that since most all modern SLR's meter wide open, any aperture selected other than the maximum is translated to the camera body mechanically via some linkage. The manufacturer is thus able to build any required exposure correction values directly into the mechanical linkage of their lens, allowing it to generate the correct meter reading for the camera as the lens is stopped down.

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Eliot, you were posting at the same time I was! Sorry for the duplication!

While I do (did) own both the Noct and the Summilux, I was not comparing them to one another. I was referring to instances with the Noct where BOTH f1.0 and f1.4 indicated a correct exposure. (FWIW, I have had the same exerience with my Summilux in low light.) This happened more than once, and usually in dim light. I thought it was an anomalie until Roberto posted his metering question a few days back -- which is what started this whole discussion in the first place. And yes, the f1.0 shot transmitted more light than the f1.4 shot, but not anywhere near a full stop worth!

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Eliot: FWIW, A) I never said anything about not liking the Noct; B) mine were not off-the-cuff remarks; C) nor were they remeberances of events from several years ago -- The last time I noticed this was less than three weeks ago; and D) if you must know, I just sold my last Noct this past week!

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

OK Jack

I take back "off the cuff". I'm sure you are relaying your experiences accurately. I have never observed the M6 meter indicating the same shutter speed for the Noctilux set at f/1.0 and f/1.4.

Some authors talk about a "T stop" (T for transmission) rather than and F stop (which relates the aperture to the focal length). I would bet that the Nocti has a T-stop of just about one stop more than the Summilux when both are at full aperture, so you do get what you pay for.

It would be interesting if someone (eg., Erwin) were to measure the T- stop of the Canon EF 50/1.0-L lens, which as far as I know is the only currently produced F/1.0 competition for the Noctilux. The Canon users might be very disappointed by the results.

-- Eliot (erosen@lij.edu), December 23, 2001.

And not to confuse the issue further, but several months back I did directly compare the Noct, the Summilux and the Summicron. For one of the tests, I shot an evenly painted wall in open shade... Interestingly, at f2, all three lenses had essentially the same amount of vignetting (slight) and all densities were uniform. At f1.4 the Noct and Summilux again produced quite similar results; a bit more vignetting than at f2, but still essentially identical densities.

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

"...and any vignetting present will affect every light ray that falls on the film, even those at the center."

Uhhh, Jack. I don't believe so.

Try the experiment I mentioned above - looking through the lens from the film's point of view. With your eye in the center of the film area you'll see a nice big round aperture at f/1 - No vignetting effect.

With your eye over near the corner you'll see an opening greatly reduced - a lot of the light rays near the corner are being blocked by optical or physical properties of the lens. It's just like using too long a lens hood, except the blockage occurs at various places throughout the lens construction and not just at the front end.

-- Andy Piper (apidens@denver.infi.net), December 23, 2001.

Andy:

If I am understanding Osterloh correctly, there are two types of vignetting, natural and artificial. I believe the effect you are describing is artificial. However at this point in this discussion if you say what I said is wrong, I believe you...

-- Jack Flesher (jbflesher@msn.com), December 23, 2001.

Jack: I think there ARE two types of vignetting, but I have no idea how they correspond to the terms artificial and natural.

Suffice it to say the Noctilux is a VERY weird lens, and - like cats and mirrors - doesn't bear too much thinking about.

Have a very good holiday, everyone.

-- Andy Piper (apidens@denver.infi.net), December 24, 2001.

I think (I could be wrong) that the difference is as follows. Artifical vignetting is the type that is avoidable, based on the design of the lens, etc. For example, if the barrel is too long or narrow, the image circle is too small for the format, the lens hood is too restrictive (based on the max aperture and focal length), or you stack filters, etc. Natural vignetting is that which is unavoidable, based on the laws of physics: ie., the fall-off of light at the corners of the frame due to the inverse square law.

Thus, to my understanding, the degree of artifical vignetting is within the control of the lens designer, but the natural vignetting is not (this depends on the angle of acceptance of the lens, the maximum aperture, and the size of the format).

-- Eliot (erosen@lij.edu), December 24, 2001.

The Noctilux vignettes A LOT at f1. But for the typical subject matter in low light it doesn't hurt the print in my experience. You just don't have to burn the edges!

-- john stockdale (jo.sto@bigpond.com), December 25, 2001.

I believe T-stops are used for movie cameras so that cinematographers can have consistency across lenses & avoid some of the problems discussed here. The 1 35mm camera system I know of that used T-stops was Bell & Howell's incredible Foton (although I'm sure there were probably a few others).

-- Chris Chen (furcafe@cris.com), December 29, 2001.