16-35mm: distortion?

[saxo]

Advertisement (gone after registration)

The technical data sheet from Leica tells nothing about distortion.

 

Distortion is of importance in wide angle lenses. Are there any data available?

 

Thanks for help!

 

Martin

 

 

[Advertisement]

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

[BlackDoc]

Have a look at the just written test from Vieri

[Bill W]

Quoting Vieri...."well controlled for distortion, showing no signs of it whatsoever". 

[Jeff S]

Perhaps corrected digitally in camera, FWIW.

 

Jeff

[MarkinVan]

Hand held, out on a walk.

 

I'm pretty happy with everything about with the lens.

 

It will be on the SL a lot

 

F3.5

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!

Edited May 4, 2018 by MarkinVan

[MarkinVan]

Looks great on my monitor, loving it.

 

F8

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!

Edited May 4, 2018 by MarkinVan

[scott kirkpatrick]

Advertisement (gone after registration)

The technical data sheet from Leica tells nothing about distortion.

 

Distortion is of importance in wide angle lenses. Are there any data available?

 

Thanks for help!

 

Martin

Geometric distortion has disappeared from the technical data sheets for "designed for digital" lenses, which will never be used on film.  Parameters for correcting it, along with lateral chromatic aberation (slightly different magnifications for different colors, at least at the edges) are incorporated in a standard way into the DNG raw file, and these corrections are also applied in camera to any JPGs and even before you see the image in the viewfinder and LCD.

 

You can read these parameters from the DNG files with some effort.  Adobe's DNG tools package has a program "dng_validate" that runs in Linux or in the terminal window on a Mac and prints out everything that is human-readable in a DNG file.

 

I have looked at a few 16-35 files this way.  A few quick and superficial conclusions based on firmware 3.1, but probably not subject to any last-minute changes):

 

The corrections are visible, but quite small at the long end (24-35mm), where the overlapping 24-90 SL Vario-Elmarit offers a wider aperture, but imposes much stronger corrections.

 

They are about 8-10X larger at the widest focal lengths (16-80 mm) but still not easily noticeable unless you need 100% image sizes.

 

There is little difference between the transformations applied to the three color planes.  Thus the need for color correction at the edges seems to have been handled almost completely by optical design.

[saxo]

Geometric distortion has disappeared from the technical data sheets for "designed for digital" lenses, which will never be used on film.  ...

 

This is a shame.

 

Digital correction of geometric distortion results in a decrease of resolution, therefore the distortion is of importance, even in lenses designed for digital.

[LD_50]

This is a shame.

 

Digital correction of geometric distortion results in a decrease of resolution, therefore the distortion is of importance, even in lenses designed for digital.

If the final output resolution is fantastic, does it matter?

 

I don’t believe it does at this point. The output of my SL lenses is great so I don’t mind what goes into it. It’s better than my similar focal length Nikkors were and better than my mostly non-corrected but similarly expensive M lenses.

Edited May 8, 2018 by LD_50

[scott kirkpatrick]

This is a shame.

 

Digital correction of geometric distortion results in a decrease of resolution, therefore the distortion is of importance, even in lenses designed for digital.

For many years barrel distortion was corrected optically by reducing the expansion at the outer parts of the frame (distortion of one form countering another).  You can see this in the distortion curves published for the fast lenses of shorter focal length that were designed up to the late 1990s.  This just produces mustache distortion, in which straight lines wiggle rather than bend.  I prefer the present approach, leaving things to software correction, which compensates exactly. 

[saxo]

For many years barrel distortion was corrected optically by reducing the expansion at the outer parts of the frame (distortion of one form countering another).  You can see this in the distortion curves published for the fast lenses of shorter focal length that were designed up to the late 1990s.  This just produces mustache distortion, in which straight lines wiggle rather than bend.  I prefer the present approach, leaving things to software correction, which compensates exactly. 

 

I agree, that the current approach of digital correction is prefered, but I still want to know at what degree digital correction is needed, as this correction aggravates image quality. I also agree, as long as you compare lenses made for digital use only, measure of distortion is not relevant, as you can measure final resolution at image edge. But distortion measure will still tell you how much digital corretion will influence image quality.

 

IMHO digital lens correction is easier (and cheaper), but inferior compared to an optical correction by a better lens design, well if possible of course. I've seen examples of digital nonsense correction: the final picture is over corrected and shows artefacts or cut real structures away.

 

Looking at the Hasselblad CFE 4/40 vs. its successor CFE 4/40IF shows perfectly the influence of distortion. The resolution of the IF version is amazing, the best I've seen in 40mm medium format (and also the mechanic is brilliant). But distortion is bad, compared to its predecessor. The argument was: easily to correct by postprocessing. This is true, but you will loose resolution, and except for the center, you end up with a minor difference in resolution compared to its predecessor.

This experience told me, that distortion of a lens is still important even in digital imaging.

 

Maybe I'm just too old fashioned.....

[saxo]

PS: the better lens design is the major reason for me to stick to Leica.

[thighslapper]

I think fairly extensive tests (by those on the forum anyway) with those current Leica lenses that have adopted FW correction of distortion have failed to show any significant reduction in image quality in the peripheries. 

 

Leica have clearly judged the degree to which they can sacrifice optical correction of distortion very well. 

 

We all tend to moan about the size of the SL lenses, but fully optically corrected ones would be even bigger and heavier. 

Edited May 10, 2018 by thighslapper

[sillbeers15]

Look at my image below, for a 16mm wide angle lens the building lines at the corner of the frame came straight out of the camera ( only corrected light intensity in LR). The building lines just look perfectly straight.

 

L1000046 by sillbeers15

[scott kirkpatrick]

I tend to believe that software correction as a final step is potentially better than optical correction for two reasons:

 

1) the limits of what optical correction can do (along with everything else that has to be managed, especially field flatness), leading to side effects like mustache distortion.

 

2) in principle, image quality should be analyzed in spatial frequency space, since that's the only way to analyze tradeoffs across the whole image plane and information at greater than the Nyquist frequency (1/the spacing between pixels) will just produce artifacts -- it isn't real when rendered in a 24 MPixel JPEG.  Yet these Leica lenses are ready to resolve at 24 MPx full frame (6 micron pixels), at 24 PMx APS-C (4.5 micron pixels), and who knows what future resolutions.  The best way to make the image fit these various rendering environments is by appropriate postprocessing.  Just as you set your sharpening for the print size.  The scene out in front of you doesn't come in nice convenient pixel sizes, but is continuous and has to be mapped into whatever chip you are currently using.

[scott kirkpatrick]

Look at my image below, for a 16mm wide angle lens the building lines at the corner of the frame came straight out of the camera ( only corrected light intensity in LR). The building lines just look perfectly straight.

 

L1000046 by sillbeers15

It's a nice picture, but LR has done some correction for you.  If you want to see the process exposed, check out this thread at getdpi:

 

https://www.getdpi.com/forum/leica/63936-images-16-35-vario-lens-sl.html#post757684 

 

 

scott

Edited May 10, 2018 by scott kirkpatrick

[scott kirkpatrick]

I did a careful comparison of the distortion corrections applied for the 16-35 in firmware releases 3.1 and 3.2.  There are some changes, which are largest at the wide end, and depend on the distance at which the lens is focused.  At very close distances, the two firmwares give the same corrections (I was at the closest possible, 7-8 cm between the lens hood and the object).  But for normal distances, say 5 m to infinity, the corrections for the 16mm focal length are different.  It looks as if they are about 10-20% weaker, and different in the details of the curve that is used.  As noted before, there is little color correction.

 

For people who think that software correction is a bad thing, this is good news -- Leica decided to use a bit less of it.  For those who think upgrading firmware is difficult and scary -- upgrade to get the most out of this lens (and perhaps others).

[FlashGordonPhotography]

 

For people who think that software correction is a bad thing, this is good news -- Leica decided to use a bit less of it.  For those who think upgrading firmware is difficult and scary -- upgrade to get the most out of this lens (and perhaps others).

 

OTOH the 3.2 firmware appears to introduce a most undesirable new "feature". And by feature I mean pain in the arse..... Compulsory image review if you don't fully disengage the shutter button between shots. What a daft idea....

 

Gordon

[Advertisement]

Hallo,

Look here: Leica L-Mount

[scott kirkpatrick]

OTOH the 3.2 firmware appears to introduce a most undesirable new "feature". And by feature I mean pain in the arse..... Compulsory image review if you don't fully disengage the shutter button between shots. What a daft idea....

 

Gordon

The review image goes away as soon as you reach half-pressed, doesn't  require a full finger lift.  But we should argue this point and get some of the really hot tempered posters involved (they know who they are).  Then it could be optional in 3.3.  But I like it this way.

[LD_50]

It doesn’t require full disengagement of the shutter release to remove the image preview. It requires you partially release the shutter.

 

The only time I think I may not releasepartually is if I’m trying to control camera shake so I’m not sure I’ll ever see this image review in real use.