16-35mm: distortion?

[caissa]

Advertisement (gone after registration)

By chance I followed some discussions in some mf fora about the distortion of the wide lenses for X1D and GFX. They found that currently there are no lenses without distortion, most have 3 to 4 % barrel type, like the new 21mm lens. Only the famous SWC biogon 38mm lens had no distortion (max. 0.2 to 0.3%). But this lens is not working well in digital scenarios. (Not very wide anymore)

Back at the SL forum I wonder why there are no similar discussions here ...   Am I too blind to find the distortion in the 16-35, or is it simply completely hidden from all users eyes by software ?

Is it possible to measure it and to tell us what the percentage is ?  (At 16mm or 18mm which is probably close to the SWC biogon lens regarding angle of view).

Edited January 17, 2019 by caissa

[Advertisement]

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

[thighslapper]

1 hour ago, caissa said:

  Am I too blind to find the distortion in the 16-35, or is it simply completely hidden from all users eyes by software ?

Correct. The codes are baked into the exif info and LR corrects it automatically.

To date no-one has shown that it materially affects the quality of the final image to any extent that is noticeable.

Leica clearly designed the lens with the correction in mind as a trade-off in correcting the other aberrations and distortions optically. As such measuring it ...... or complaining about it is fruitless. If I recall correctly some RAW processors such as C1 allow you to switch off the corrections and view the original image if you wish. 

Perfection in WA zooms is almost impossible to achieve optically..... and getting closer would come with a sizeable increase in size/weight and cost.

Edited January 17, 2019 by thighslapper

[scott kirkpatrick]

Herr Schenkelschlager (Thighslapper) has it right. Designing zoom lenses and very wide angle lenses is hard, but today we can correct lenses at several stages of the digital pipeline.  Barrel/pincushion distortion and chromatic aberrations are more easily corrected in software than optically.  Optical corrections applied in classic lens designs often left behind higher order errors like "moustache distortion," which can be worse.  Leaving them in the optical path to be resolved during rendering allows the designer to correct other things.  Sean Reid writes repeatedly about the possible loss of fine scale resolution due to expanding the barrel-distorted image which is captured to a larger pincushion-shaped image which correctly reflects what the lens should have seen.  This is then cropped to the 24 MPx that we paid for as a final image.  I have never seen in Sean's examples, where he leaves out the corrections by a complicated multi-step process, anything that I would find very troubling.  But if you want to see what is being done, read Sean's descriptions of his methodology for LR, or just work in Capture One and go to the "lens profile" (Versions <=11) or "lens correction" (v12) window, uncheck "hide distortion corrections," and turn the correction down to 0%.   Or use a rendering software that doesn't apply the corrections like AccuRaw, by Sandy McGuffog.

I think the bottom line is that the lens designers know why they are doing this, now that all lenses are designed for digital, and overriding it won't lead to the most accurate images.  I can even argue that worrying about the crispest possible gravel in the corners of an image is enhancing artifacts, not true image data, but that takes some signal processing jargon, and I'll keep that to myself.

[helged]

For what it is worth - for one particular application: When testing several wide angle lenses for astrophoto, the lens with best sharpness across the frame and essentially with no coma etc. towards the edges/corners, was the SL24-90mm at 24mm (@f2.8). Actually, the 24-90 did a better job than the otherwise excellent Zeiss Otus 28mm (at f1.4 and f2.8; tested both on the Nikon D810 and SL). I have not tested the 16-35mm for astrophotography yet (using the Sigma 14mm f1.8 Art for the time being), but I expect the performance to be very, very fine. Yes, the end result is a mix of optics and software correction(s). The software corrections dont bother me at all - as long as the end result is excellent. For my eyes, of course...

[jaapv]

6 hours ago, caissa said:

By chance I followed some discussions in some mf fora about the distortion of the wide lenses for X1D and GFX. They found that currently there are no lenses without distortion, most have 3 to 4 % barrel type, like the new 21mm lens. Only the famous SWC biogon 38mm lens had no distortion (max. 0.2 to 0.3%). But this lens is not working well in digital scenarios. (Not very wide anymore)

Back at the SL forum I wonder why there are no similar discussions here ...   Am I too blind to find the distortion in the 16-35, or is it simply completely hidden from all users eyes by software ?

Is it possible to measure it and to tell us what the percentage is ?  (At 16mm or 18mm which is probably close to the SWC biogon lens regarding angle of view).

Actually we see a new way of designing lenses. Correcting optical aberrations means shifting  and compensating them by adding lens elements. With the new hybrid designs the optical part of the lens is corrected better than was possible in the past by shifting the residual aberrations into distortion. Then the distortion is corrected digitally, leaving a higher image quality than before. Another benefit is that the distortion can be reduced to something approaching 0.

So yes, it is "hidden" digitally from the eyes of the users - by superior design.

[scott kirkpatrick]

8 hours ago, caissa said:

.... Only the famous SWC biogon 38mm lens had no distortion (max. 0.2 to 0.3%). But this lens is not working well in digital scenarios. (Not very wide anymore)

Just on this one point:  I have an SWC and use it with the largest of the Phase One CCD backs, the P45+, which is big (48x36 mm) but only 39 MPx.  Still, the Biogon gives me roughly a 25mm-eff field of view.  Some of the IQ CCD backs have more MPx but are smaller (44x33 mm.).  Only the latest 100 MPX or better are reaching the full 60x45 mm size and none cover the 58x58 mm of 120 film.  At Bear Imaging they told me that the latest, biggest IQ backs do not attach to the early Hasselblad V-series because those lenses don't resolve well enough and the old cameras cannot be aligned well enough to get the best out of a 100 MPx sensor,.  But the SWC and its Biogon give me lovely results with the older back.

[Peter Branch]

Advertisement (gone after registration)

The first published reference of which I'm aware to using software in conjunction with optical design to give an overall optimal performance was an Eastman Kodak Company Annual Report from about 35 years ago!   In that instance it was suggested that "Satisfactory" performance could be realised with very much cheaper lenses.  Specifically it was claimed that software was a superior way of correcting distortion.

What somewhat surprises me is that it has taken this long to be realised.