Correcting Distortion

[Samouraï]

What is the general consensus concerning computer correction of optical distortion? I have always tended to resist post-correction and have preferred shooting with a 50mm lens on a 135 system for that reason. Now that I see the kind of distortion that occurs at the edges of a frame with a 50 on a 6x7 system, I find myself not liking the 50 as much as I thought I might. I still like the way the 50 treats the z-axis, though.

I suppose the increased bending of light, and thus distortion, can't be helped when the format is so much larger, or is this not true.

 

[segedi]

I think it depends on what camera/lens you use, the[FONT=verdana, arial, helvetica] Mamiya RZ 50mm f/4.5 ULD L is known for not having much distortion at all.
[/FONT]

 

[Samouraï]

I've read up on the Mamiya 6 50mm lens. And I'll get it someday soon.

So in a perfect world, when does distortion enter into a lens' natural capturing of a subject? As 50mm is considered to treat space in the same way/in a very similar way that the human eye captures space (not talking about fov), should distortion naturally be exhibited in any lens with a focal length less than 50mm?

And as an aside, what lens actually captures reality in the most accurate way? What lens might one use to project an image so that nothing is distorted? I have heard a 40mm lens is often used for blowups...

 

[sevo]

So in a perfect world, when does distortion enter into a lens' natural capturing of a subject? A

There is no "natural capturing" - every lens projects a three-dimensional structure into a two-dimensional plane, and there are quite a few possible projections that may deliver a natural impression when enlarged to the right scale and viewed at the right distance.

We can define a lens as "distorting" when it deviates from the ideal projection it is intended to have - but that's it. Most people that complain about "wide angle distortion" don't actually mean the projection error of a given lens, but feel disturbed by the unnatural look of many wide-angle shots viewed at too small enlargement or too large distance...

 

[Samouraï]

I understand how lenses work. I guess I was asking the wrong question. I was basing my standards on the human eye and a straight line. I meant to question the curvature of a straight line in wide angle lenses. I know, it's kind of a sophomoric to be asking questions and assuming standards. And it may be a silly question, but I am attempting to understand something slightly more complex than what I am asking. Scientific ideals vs real world representations of light bending and all that.

And by the way, my wording: "natural capturing" was supposed to mean "ideal projection."

Thanks for the explanation, though, and I think it does help.

 

[Samouraï]

And this began with the idea of distortion correction in post-processing applications like Photoshop. And I've seen examples of 18mm photographs of architecture being "distortion corrected" to offer straight lines and a level perspective. I am assuming this is a misuse of the term distortion, and rather something more along the lines of perspective correction?

 

[Tim Gray]

When I've seen examples like that, they could be referring to either actually lens distortion (barrel, pincushion, wave/complex) or perspective distortion. It really depends on the picture.

 

[Samouraï]

gotcha.

but all the same, at what focal length do straight lines begin to bend in an ideal spherical lens? 49mm? 39mm? maybe that's my question. and i guess the second variable is frame size, or does this not affect the outcome when imagining an ideal spherical lens?

if one had an infinitely large focal plane and its necessarily-sized, ideal 50mm lens, would it exhibit the uniform properties of a 50mm lens across the entire image? or would there be marked bending of straight lines beyond a certain image size (assuming a 50mm lens can create uniform straight lines in 135 and medium format, regardless of subject distance).

now, i don't mean to annoy anyone. i am just wondering if someone out there can understand what i am trying to ask, and lend some knowledge/wisdom. thanks.

 

[Tim Gray]

So if you mean at what focal lengths are straight lines no longer straight and NOT some kind of perspective distortion, the answer is a bit more complex. Theoretically, it's up to lens design and not focal length dependent. One could imagine a lens design that has no distortion like that at any given focal length. Practically, especially on SLRs, you tend to get more complex distortion at wide focal lengths. But if you can use any lens design you want, you can probably avoid distortion at just about any focal length.

Since this is RFF and I'm more familiar with M lenses, I'll name just a few lenses at different focal lengths which have essentially no distortion:

- Leica 50mm Summicron (current version)
- Zeiss Biogon 35mm f/2
- Leica 28mm f/2.8 Elmarit ASPH
- Zeiss Biogon-C 21mm f/4.5
- Contax G Hologon 16mm f/8

It's generally easier to design telephotos without complex distortion, so there's a bunch at longer focal lengths with relatively little distortion or just mild pincushion distortion.

On the other hand, if you are still talking about perspective distortion, well that's not really a function of focal length, but of camera to subject distance (and viewing distance of the resulting image). It is more evident with wide angle lenses because of their angle of view. Most people don't object to perspective distortion around the 35mm focal length. Some don't like 28mm because it can have a bit too much in close situations, though I find that a focal length that is pretty benign with respect to perspective distortion if limited to 0.7m or more. 24mm and 21mm (and certainly anything wider) can give you perspective distortion pretty easily in most situations.