"digital" M lenses

[jankap]

Advertisement (gone after registration)

I can see the difference that 0.01mm makes when setting the shim for a 5cm F1.5 Sonnar, using the M9 or M Monochrom for the final adjustment. I cannot focus any better than that. The thinnest shim used in the 1930s Sonnar was ~0.02mm, after that- combine shims and use the variable stand-off ring. For film, getting it within 0.04mm was good enough.

 

Could you please tell more in detail, what you have done with the Sonnar?

It interests me, what is possible.

You can add distance somewhere in as lens to optimize it. But it is not possible to add negative shims, isn´t it.

 

I think, that sensors can deliver sharper pictures. I suppose, that, if the sharp picture is 0.01mm under the film surface, the IQ suffers.

Jan

[Lenshacker]

Could you please tell more in detail, what you have done with the Sonnar?

It interests me, what is possible.

You can add distance somewhere in as lens to optimize it. But it is not possible to add negative shims, isn´t it.

 

I think, that sensors can deliver sharper pictures. I suppose, that, if the sharp picture is 0.01mm under the film surface, the IQ suffers.

Jan

 

Usually the lenses have shims in them, and can usually substitute thinner ones. I've also polished down the variable retaining rings of the Sonnars.

 

RIMG0048

 

Jason Howe hosts the instructions here:

 

50mm Jupiter 3 f/1.5 Information - Aperture Priority

 

You can convert the pre-war Carl Zeiss Jena 5cm F1.5 and F2 Sonnars to Leica mount using the focus mount from the J-3 and J-8, respectively.

[adan]

On the original question - It is complicated.

 

I would say that the 35 f/1.4 FLE is an example of a lens being redesigned to reduce focus shift, which is more critical with the "infinitely thin" focus plane of digital. The previous design got away with some focus shift (and few complaints) for 15 years due to the "depth" of film emulsions.

 

I would say that the manufacturing tolerances have been tightened - but that is not really a question of "design". Rather, building a given design more precisely to spec.

 

It is quite possible that the newer wide-angle lenses have been designed with digital sensor architecture in mind. Definitely the pre-2000 wide-angles need a LOT of in-camera processing to correct for their steep angle of incidence.

 

The way to test this would be to turn off lens recognition in an M240 or M9, and then compare the "raw" unprocessed performance of pre-digital 21 Elmarits with the more recent 21 Elmar, 21 Summilux, and Super-Tri-Elmar set to "21mm".

 

If the newer lenses show less overall vignetting, less color vignetting (green/cyan corners from IR filtering) and less "Italian-flag" color shading - with no in-camera corrective processing - then it would be clear that the newer designs are intentionally more "digital-friendly" than the pre-digital designs.

 

6-bit coding is obviously only needed for digital. But since most lenses back to 1980 can be coded, they are all potentially "digital" lenses as well as "film" lenses, in that regard. I happily use c. 1981 "Mandler" 21 and 35 lenses on digital. Both are fuzzy in the corners wide open - but equally fuzzy on either film or digital.

[Lenshacker]

I just picked up a 65 year old Minolta 8.5cm F2.8 in Leica Thread Mount- "threw it on the M Monochrom". The focus is perfect from minimum focus of 4ft to infinity. The same is true of the 1940s and 1950s Nikkors that I have- focus is perfect throughout the range. They must have assumed that film was perfectly flat in the camera, it certainly works out for Digital.

[Exodies]

Is it possible to design a lens with improved image quality by reducing the optical correction for one particular aberration (on the grounds that that fault can be satisfactorily corrected in software) thereby obtaining more leeway for correcting another deficiency?

If so, this would create a class of digital lenses as they would be worse without software correction.

[adan]

Possible, yes.

 

ALL lens designs are a compromise trading off more of one kind of optical problem to improve something else (e.g. Noctilux f/1.0 - accepted substantial vignetting for improved corner clarity).

 

A lot of recent lenses have been designed with optical problems left in, because software can correct the problem. Most commonly barrel/pincushion distortion, but also color aberrations, etc..

 

But to be frank, this is usually done as much to correct the "deficiency" of HIGHER COST, rather than for better performance somewhere else. Save money designing and building an imperfect lens, on the assumption that software can correct the imperfections more cheaply than better design, or more expensive materials. Assuming the imperfections are of a mathemically "computable" kind - which most are.

 

Cost, of course, is not Leica's row to hoe.

[gpwhite]

Advertisement (gone after registration)

Why "older spec"? Your 28/2 is 6bit coded and calibrated for digital if it was assembled after the M8 i guess.

 

The 28 f/2 precedes the Karbe Era, so I think of it as older spec.

 

Look, my best images are with the APO 50, so clearly I admire Herr Karbe's lens genius. But I am a wide-angle guy, and the wide Karbe Summiluxes leave me cold. Other than the 18 SEM, which is 2009-ish design, my three preferred wide-angle lenses are older spec (19/2.8, 24/2.8 and now 28/2).

 

But the thread asked whether there are digital-friendly lenses, and my comment was (just like Jaap's opinion that followed) that I think any Leica lens can be calibrated to perform as well on digital M as film M.

[Hey You]

There is a concave rear lens element on each of the 50 APO, the 35 Summicron ASPH and the 90 Macro Elmar.

Other lenses also have such an element, but these are the three I am aware of.

Does use of a concave instead of a convex rear lens element have anything to do with digital?

[gpwhite]

There is a concave rear lens element on each of the 50 APO, the 35 Summicron ASPH and the 90 Macro Elmar.

Other lenses also have such an element, but these are the three I am aware of.

Does use of a concave instead of a convex rear lens element have anything to do with digital?

 

The Summilux-M 35mm ASPH design, which long predates the current FLE offering, has a lovely concave rear element. I got mine in 1996, so there is no digital sensor correlation with that model. I imagine there are others too, but I just remember looking at that back end of that lens

[adan]

Does use of a concave instead of a convex rear lens element have anything to do with digital?

 

No. As mentioned, the 35 f/1.4 Aspherical AND the 35 f/1.4 ASPH had concave rear elements as early as 1990/1994. (And for that matter, the 135 Elmarit-M first design from 1963!) Long before digital was of any importance. It is just a tool for the lens designer to use when appropriate, regardless of what the imaging material is.

 

In the case of the 35s, the Leica designer essentially took the classis "double-gauss" 6/7-element design, and gave it a pair of negative (concave) elements front and back, to act as "spectacles" to improve on the limitations of the DG formula - brilliant design thinking!

 

BUT -

 

If anything, it's my own opinion that concave rear elements are a BAD idea for use on digital - no matter how helpful they are to overall image quality. The concave shape acts like a mirror lens, focusing any light that reflects back from the imaging surface (film or sensor). And since sensors are much more reflective than film - light bounces off the sensor, hits the concave rear element, and is reflected back onto the sensor as a semi-focused bright central patch of fogging.

 

(A convex rear element will also reflect light, but scatter it rather than focus it. And film is too "matte" to reflect as much light as a sensor.)

 

I've run across this with both the original 35 ASPH Summilux and the 75 Summicron (also concave on the back) in some situations, usually a bright central subject against a dark background. The flare even picks up the shape of the subject sometimes (flipped upside down and sideways, showing it is an optical reflection, not some other kind of flare).

 

They are amazing designs for film - but I don't use them for digital any more.

[tobey bilek]

RF focus spec is the same, i.e. nominal, but the tolerence has been made smaller.

An older lens might have been fine depending on where in the tolerance range it fell.

 

I recently purchased a new 28 2.8 ASPH and the instructions said it was the first lens designed for digital. Having used a few older ones, I can attest to the very far corners are superior. The longer the focal length, the less it seems to matter.

 

Nikons are the same way, but perhaps for different reasons. 24 and shorter and some 28 are terrible in the far corners, but they work well on film so it is not the lens being defective. Nikon is not so free with details of sensor design.

[CheshireCat]

I think the most important reason why the latest wide angles are retrofocus-like designs is to work around issues with the angle of incidence on digital sensors.

 

Also, I believe that the latest lenses hasve improved internal reflection countermeasures to minimize issues with rays being reflected by the sensor surface.

 

Therefore, yes, "digital-aware" lenses do exist.

[semi-ambivalent]

the angle of incidence on digital sensors.

 

I had never thought of this. In my original post I was actually thinking of curvature of the focused image at the plane of focus. I thought maybe film emulsions were thick enough that some of the aerial image could come to focus inside the emulsion or at various depths within it; a luxury not available in digital, making the (digital's) OOF more visible, especially with today's high pixel counts and print sizes or serious cropping.

 

thanks,

s-a