Terrible color aberrations with older Elmarit 21mm

[Tulpenwahn]

Advertisement (gone after registration)

Dear Forum,

For years I only used a Summilux 50mm ASPH on my M9 and never had any colour aberrations or chromatic aberrations worth mentioning. Now I have bought a second-hand Elmarit 21mm (from 1995) and the result is sobering: blatant violet colour fringing on the lower left edge of the image. What am I doing wrong or what can be done about it?

I actually thought these problems had been solved sometime in 2008/2009 (at least I can remember such discussions in the dark). As I said, there is no problem with the Summilux.

Kindly Winfried

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[lct]

Just color shifts due to the use of pre-asph WA lenses on early digital sensors. What you did wrong is perhaps forgetting to select this lens through the Lens Detection menu. If this does not work, using Cornerfix should fix the problem.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[AndreasG]

vor 7 Stunden schrieb lct:

early digital sensors.

Yes, with later sensors (M10) no problem at all, I use a 1990 21mm Elmarit without any colour fringing. And I do not select the lens since I use a M10-D without display and this function.

[adan]

The 21mm Elmarit-M works fine on the M9 - IF IT IS 6-BIT CODED. (Or correctly selected in the camera's lens menu).

If not correctly identified for the camera, yes, the purple/pink-colored edges are to be expected. And snow is the perfect subject to reveal them.

My 21mm Elmarit was the very first lens I sent to Leica for coding (in 2007, for use on the digital M8).

It is not really a color aberration - it is just an interaction between wide-angle Leica rangefinder lenses and the microscopic structure of digital sensors.

It is exactly the reason Leica had to introduce the 6-bit coding on lenses in the first place (especially wide-angles). And indeed the problem was solved in 2006 - 6-bit coding was the solution (to allow the camera to automatically neutralize the pink pattern for each different focal length).

BTW - is that first picture taken on the Teufelsberg (Spione-Amt)?

Edited February 11, 2021 by adan

[Tom1234]

QUESTION: So in summary,

Was the purple or pink cast fixed on new sensors thus NOT needing software correction? 

Or do you still have to identify the lens or a similar millimeter lens?

Or is it somewhere in between, like less color fringing even without identifying the lens, but it is still best to identify the millimeter of the lens in the camera menus.  

[adan]

8 hours ago, Tom1234 said:

.......somewhere in between, like less color fringing even without identifying the lens, but it is still best to identify the millimeter of the lens in the camera menus.

Correct answer. The M10 sensor (and I guess the M(240) sensor - I skipped that generation) are a bit more tolerant of wide-angles.

But not really enough for me - I still see "color artifacts" with most uncoded lenses 35mm and wider.

It has to do with two improvements. Reshaped (taller) microlenses over each pixel, and a shallower 3D structure in the silicon itself, that reduces "leakage" of, say, red-filtered light from a "red" pixel into a neighboring green-output pixel, or vice versa (or blue onto green and vice versa).

And technically, "similar mm" is not enough. The actual optical design has to be similar. The same coding will not work perfectly (or even at all) for: a 21 f/2.8 Elmarit from 1980, a 21 f/3.4 Super-Elmar ASPH from 2011, or a 21 f/3.4 Super-Angulon from 1970 (there is no coding at all that will work with the ancient Super-Angulon). It is why each specific model of lens has its own coding assigned.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Tom1234]

Advertisement (gone after registration)

1 hour ago, adan said:

Correct answer. The M10 sensor (and I guess the M(240) sensor - I skipped that generation) are a bit more tolerant of wide-angles.

But not really enough for me - I still see "color artifacts" with most uncoded lenses 35mm and wider.

It has to do with two improvements. Reshaped (taller) microlenses over each pixel, and a shallower 3D structure in the silicon itself, that reduces "leakage" of, say, red-filtered light from a "red" pixel into a neighboring green-output pixel, or vice versa (or blue onto green and vice versa).

And technically, "similar mm" is not enough. The actual optical design has to be similar. The same coding will not work perfectly (or even at all) for: a 21 f/2.8 Elmarit from 1980, a 21 f/3.4 Super-Elmar ASPH from 2011, or a 21 f/3.4 Super-Angulon from 1970 (there is no coding at all that will work with the ancient Super-Angulon). It is why each specific model of lens has its own coding assigned.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Thank you... this is exactly what I needed to know.  

I have that SA 21mm so it is out of luck with the new sensors.  I also have a 35mm 1.4  V4 a small lens with most angular approach rays probably not coded.  But I assume the other lenses have some kind of correction circuit?  So what to do with previous NON Coded lenses? 

There is a list of lenses that won't work with the M9 & M8 maybe that list (21mm 3.4 SA and certain earlier 28mm F2.8 by serial number) are the only lenses that have no correction program?  I will have to download the SL2-S manual to see the list of lenses that they have corrected for. 

But on the SL2-S one would hope all will work given the new L-Mount but not so if lens correction selections must still be made in the menus.  I have over 18 old lenses.  

Yet maybe I do not want vignetting corrected since it can be used creatively.  I am not into the "perfect" look of the modern lenses. 

[Tom1234]

29 minutes ago, Tom1234 said:

I have that SA 21mm so it is out of luck with the new sensors.  I also have a 35mm 1.4  V4 a small lens with most angular approach rays probably not coded.  But I assume the other lenses have some kind of correction circuit?  So what to do with previous NON Coded lenses? 

There is a list of lenses that won't work with the M9 & M8 maybe that list (21mm 3.4 SA and certain earlier 28mm F2.8 by serial number) are the only lenses that have no correction program?  I will have to download the SL2-S manual to see the list of lenses that they have corrected for. 

But on the SL2-S one would hope all will work given the new L-Mount but not so if lens correction selections must still be made in the menus.  I have over 18 old lenses.  

Yet maybe I do not want vignetting corrected since it can be used creatively.  I am not into the "perfect" look of the modern lenses. 

Found the coded lens list here: https://lavidaleica.com/content/leica-lens-codes#rlens

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

M Menu selectable lenses on some previous Leica M cameras.

Accuracy of this list for the sl2-s unknown

R Menu selectable lenses on some previous Leica M cameras.

Accuracy of this list for the sl2-s unknown

 

Encoding your own lens: https://bophoto.typepad.com/bophoto/2009/01/m8-coder-simple-manual-handcoding-of-m-lenses.html

Edited February 23, 2021 by Tom1234
Deletion unneeded material.

[adan]

5 hours ago, Tom1234 said:

I have that SA 21mm so it is out of luck with the new sensors.  I also have a 35mm 1.4  V4 a small lens with most angular approach rays probably not coded.  But I assume the other lenses have some kind of correction circuit?  So what to do with previous NON Coded lenses?

Some lenses that cannot have coding installed (because Leica changed the physical chrome lens ring from 2-3mm thick to 1mm thick in the 1970s) can be identified by a camera menu.

That includes the 35mm f/1.4 pre-ASPH (any version), as well as the 135 f/4 Tele-Elmar.

The 21 Super-Angulons do not appear in the menus (too old, wrong mount type) but can and have been used successfully for digital B&W pictures, either on the Leica Monochrom M cameras, or used on the full-color cameras and converted to B&W. However they still block the TTL metering somewhat, just as they did with the M5 or M6, and need to be light-metered manually. (on an SL/SL2, the off-the-sensor metering should work normally)

Below is a picture I made with the 21mm Super-Angulon f/3.4 on an M9. Both the original full-color image (with pink stains), and desaturated to B&W.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

[Tom1234]

"Dirty" Ipod pictures of my 21mm 3.4 SA.  

Mine is a later version with ground down retaining ring to provide the camera meter light.  I was told when purchasing used off Ebay that this was a DAG Repair Shop modification, but he denied it on the phone, and I later read on the internet that Leica did this on later lenses to provide some metering ability, though not accurate.  

Metering with the 21mm 3.4 SA in my experience is too bright when compared to a 24mm 2.8 ASPH.   Apparently the lens element being too close to the meter causes the "too bright" error. 

On M7:    Infinity focus 2 stops too bright.    Mid focus assumed 1.5 stops too bright.    Close focus 1 stop too bright. 

On M9:    Infinity focus 3 stops too bright.    Mid focus assumed 2.5 stops too bright.    Close focus 2 stops too bright. 

 

See pdf below

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

See this PDF of the lens:

21mm sa pictures of-print150Highj2k.pdf

 

Edited February 23, 2021 by Tom1234
Clarifications.

[Jean-Michel]

You can correct the “Italian flag,” issue with either the Lightroom flat field plug in or by using Cornerfix.