RF lens distortion v DSLR

[gravityassault]

Hi everyone,

I hope you're all doing well. I have a simple question: why is it that my Zeiss 25 Bigon on an M6 seem to give far less distortion than a similar focal length, (mid-range on a 17-35mm lens), mounted on a Nikon D2X?

Your thoughts are most appreciated.

All the best,

Chris

 

[JRG]

Well, the Biogon is a true wideangle lens, not a retrofocus design. Also, the 25 Biogon is claimed to be one of the most distortion-free lenses in production.

 

[degruyl]

I suspect that you would find less distortion on any RF (or view camera) lens in the "wide angle" range.

The fact of the matter is that these lenses don't need to do acrobatics (retrofocus) to allow for a swinging mirror. In other words JRG is correct.

 

[gravityassault]

Thanks

Thanks

Thanks guys for your thoughts. Is there a physical reason behind the lower distortion? Is the glass different? Does the distance from the front and rear elements make a difference between RF and SLR? What causes the images to have low distortion?

Chris

 

[ampguy]

If you want a very low distortion lens for your SLR, and can adapt M42 mounts, look for the Sigma 18mm/3.5. I couldn't tell the difference between it, and Summicron 35 photos taken further back.

For RF value, the CV 15, and 21 are great values, but both have more distortion than the Sigma mentioned above.

 

[Arjay]

Is there a physical reason behind the lower distortion? Is the glass different? Does the distance from the front and rear elements make a difference between RF and SLR? What causes the images to have low distortion?

Inherently, wide-angle lenses for RF cameras have their rear lenses fairly close to the film plane.

Imagine the film image window, and some hypothetical lines that describe the outermost light rays to the edges of the film window. The angle between these lines is defined by the lenses viewing angle. By design, these lines need to intersect in the lenses optical center. The shorter the focal length of the lens (the wider the viewing angle), the closer the intersection of these lines will move to the film plane, which explains why the rear lens will extend particularly close to the film plane.

SLRs need to have a minimum distance between the lenses rear element and the film to provide for the motion of the viewfinder mirror. This distance is a serious design problem for SLR wideangle lenses, as inherently these would need to extend much further into the camera body than is permitted for a meaningful VF mirror motion.

On order to provide space for this mirror, wideangle lenses for SLR cameras need to use a specific design that moves the entire lens assembly further away from the film plane, effectively featuring different viewing angles for the front and back (film) sides of the lens.

In order to achieve this goal, SLR wideangle lenses use a so-called retrofocus design, which includes a fairly large number of additional lenses over the equivalent RF camera lens. These lenses are what introduces the additional distortion in SLR objectives.

You can see this conceptual diffference between RF cameras like the M-bayonet Leica and any SLR in the distance between lens bayonet flange and film plane. Virtually all full-format and APS-C format SLRs feature a film-to-flange distance that is almost double that of the Leica M bayonet flange to film distance.

As a consequence, it is hardly possible to adapt any RF lens to an SLR camera without permanently moving the VF mrror out of the way - effectively disabling the SLR's viewfinder and its associated functions such as autofocus and exposure measurement.

 

[Darshan]

Inherently, wide-angle lenses for RF cameras have their rear lenses fairly close to the film plane.

Imagine the film image window, and some hypothetical lines that describe the outermost light rays to the edges of the film window. The angle between these lines is defined by the lenses viewing angle. By design, these lines need to intersect in the lenses optical center. The shorter the focal length of the lens (the wider the viewing angle), the closer the intersection of these lines will move to the film plane, which explains why the rear lens will extend particularly close to the film plane.

SLRs need to have a minimum distance between the lenses rear element and the film to provide for the motion of the viewfinder mirror. This distance is a serious design problem for SLR wideangle lenses, as inherently these would need to extend much further into the camera body than is permitted for a meaningful VF mirror motion.

On order to provide space for this mirror, wideangle lenses for SLR cameras need to use a specific design that moves the entire lens assembly further away from the film plane, effectively featuring different viewing angles for the front and back (film) sides of the lens.

In order to achieve this goal, SLR wideangle lenses use a so-called retrofocus design, which includes a fairly large number of additional lenses over the equivalent RF camera lens. These lenses are what introduces the additional distortion in SLR lenses.

Awesome explanation. Thx.

Dan.

 

[pvdhaar]

The explanation may be awesome, but is not true..

The real reason that some lenses have barrel distortion, and others don't, is a design choice by the manufacturer. You'll find that Canon and Nikon can build perfectly rectilinear wide angle lenses for their SLRs; the perspective control lenses are examples of this.

To achieve this, the magnification has to be non-uniform over the field. The more off-axis, the more magnification is required to avoid converging lines. You want a brick wall to look straight in the picture, which in reality it doesn't; parallel lines converge in the distance, and if you had uniform magnification over the field of the lens, they'd do that in the photo too.

The downside to perfect rectilinearity is that because of the marked unevenness in magnification off-axis, circular objects appear as egg-shapes when not in the center of the frame. And that's not what the average photog wants to have happen to people along the edge of the picture. That's why some amount of barrel distortion in wide angles is desirable.. it keeps heads at the sides more or less round.

Canon and Nikon know this, and build their general purpose wide angles with a moderate amount of barrel distortion, and their PC lenses rectilinear..

 

[Arjay]

The explanation may be awesome, but is not true.

Can you please explain this in greater detail - your post doesn't explain it either.

Your explanation of distortions as non-uniform magnification which is a function of an image area's distance from the optical axis is just a more accurate description of what distortion is, but doesn't give any clue of what causes this distortion/non-uniform magnification.

Admitted - the simple RF-type lens designs can also exhibit distortion under extreme conditions, but they are by far not as serious as with normal SLR designs.

The cause for the larger distortion in SLR lenses is predominantly their retrofocus design. If you can explain this in greater detail or prove that I am wrong, then please do.

 

[sevo]

Canon and Nikon know this, and build their general purpose wide angles with a moderate amount of barrel distortion, and their PC lenses rectilinear..

You are right in that lenses do get designed to have residual barrel rather than pincushion distortion as the former errs towards the perceptively natural side. But that hardly means that they intentionally add distortion where they could avoid it without extra effort - they certainly have not done so on (symmetric and hence inherently non-distorting) rangefinder lenses.

SLR lenses simply are more complex as a consequence of requiring a retrofocus construction, and a perfect rectilinear lens under that restriction is expensive in design and construction - too expensive for a consumer market likely more bothered by the egg-shaped projection of peripheral heads than by non-rectilinear architecture and product shots.

Sevo

 

[squinza]

The most symmetrical a lens is, the less distortion it has, that's a general rule. Think, for example, to the hypergon.
SLR wide angle lenses need a retrofocus design, as Arjay has explained very well, so they are not symmetric at all, that's why as a rule SLR lenses show more distortion, especially with wide angles.
Anyway, modern design techniques made possible the making of excellent, distortion free wide angle lenses. Think of the new TS-E 24mm by Canon: the mark I version of the lens (which I have) shows visible distortion. The new version is reported as being perfect...

 

[pvdhaar]

I'll give an example..

Think about it this way:

You're standing in front of the great wall of China in the middle of a flat plane. The masons have built it perfect, so you know it's exactly straight at the top for miles and miles to the left and right. But...

Because objects further away appear smaller, in your left and right peripheral vision you notice that the top and bottom of the wall converge far left and far right. So, what you see is that the top and bottom of the wall touch each other far to the left and to the right. Which of course they actually don't, but that's what you see if your field of vision were wide enough..

If you do not want to see the wall tapering away in the distance, you have to fight the principle that objects further away appear smaller. So, at the periphery of the field of view, they have to be rendered larger. In short; off-axis, the magnification has to go up.

If the magnification off-axis holds equal thread with the smaller appearance of objects further away to the left and right, then what you get is a rectilinear lens. It renders flat surfaces flat and straight. Great for shots of brick walls. The price you pay however is that surfaces that are not flat and straight are handled the same way, and appear distorted from what your eye/mind are used to see.

Your eye is built such that if you turn your head, and the field of vision changes, the shape and size of objects that you see remains the same. The magnification is more or less uniform in your entire field of view.. But if your field of view would be large enough, you'd notice the barrel distortion when looking at things like the great wall of China. An example of an extremely wide lens where the magnification is uniform across the frame is a fisheye.

Lens designers have to make a choice where between these two extremes. The purpose of a lens decides what the choice has to be.. Architecture or documentary, or cramming everything in the field of view.

Obviously, getting a lens perfectly rectilinear is not trivial, but with modern (computer assisted) lens design techniques is perfectly possible even for an SLR that needs a retrofocal approach to accomodate the mirrorbox.

 

[Arjay]

That's a great explanation of what is rectilinear image mapping. But if I'm reading this thread correctly, that's not the topic of this thread.

The topic is why is making a rectlinear lens for SLRs more difficult than making such a lens for RFs, or why exactly does a rectilinear lens design introduce distortion in how such a lens maps its image.

I am very well aware that physiologically, our eye's vision does also exhibit some severe distortion - we just don't perceive this, because there's a beautifully powerful image processor behind our eyes.

The very differences between our perceived vision and how our cameras are imaging the world makes photography such a challenge.

 

[pvdhaar]

That's a great explanation of what is rectilinear image mapping. But if I'm reading this thread correctly, that's not the topic of this thread.

No, it explains that the amount of barrel distortion is a design choice, why there are good reasons to go one way or another in that design choice, and that therefore it is not governed by a lens being retrofocal or not..

 

[Arjay]

I'm sorry. I had taken offence in your saying my explanation was wrong - but I now realize we're talking at crossed purposes.

 

[pvdhaar]

I'm sorry. I had taken offence in your saying my explanation was wrong - but I now realize we're talking at crossed purposes.

I think we've approached this from different perspectives.. I could have worded my comment differently as well; sorry for that!

I just took it from an engineer's point of view; there's no fundamental impossibility, and didn't first consider it from the practical side: retrofocal is more complex (expensive) to get it right..

 

[flip]

Silly question to add to this discussion....
Given the closer distance from lens to film plane, is it safe to say that Olympus Pen camera lenses can be more easily made at low distortion levels than larger SLRs' lenses?

 

[Arjay]

I just took it from an engineer's point of view;

We're colleagues, and probably both tenacious ... 😉

 

[sevo]

Silly question to add to this discussion....
Given the closer distance from lens to film plane, is it safe to say that Olympus Pen camera lenses can be more easily made at low distortion levels than larger SLRs' lenses?

Sort of - however not due to lens/film distance, but due to the difference between linear and area rule that govern the focal length/film size ratio.

Sevo

 

[Steve_F]

And of course software like Nikons own Capture NX2 can recognise the Nikon lens used and apply automatic corrections to distortions and aberrations.
Not related to the original post, as such, but it must make the lens engineers job that little easier knowing the software guys can 'polish-up' the final image.
It's not an excuse but if the lens guys get it within acceptable levels then job done?

Since Ive been using prime Leica I've really noticed any minor distortions in my Nikkors. It really stands out now.

Steve.