Differences in the Distortion of Facial Features

[michael_dakin]

I understand why field-of-view-wise a 35mm lens on 1.5x-crop-factor DSLR is roughly equivalent to a

50mm lens on a 35mm SLR (or full-frame DSLR). People often say a the 35mm lens on the 1.5xDSLR is

"equivalent" to the 50mm lens on the full frame. This logic leads to lens recommendations based solely

on achieving an equivalent field of view between the 1.5xDSLR and the full-frame.

 

It seems to me that there might be more to this however than field of view. In most cases the look is

what is important not how far one stands from the subject to get the picture.

 

I'm curious about how the 50mm lens on the SLR distorts facial features relative to the way the 35mm

lens on the DSLR distorts facial features when the same picture is composed of the same face with both

cameras/lenses.

 

For example say I use the SLR/50mm to fill the frame with a given person's face. At such a close

distance the 50mm will tend to distort the face of the person in subtle ways, maybe flatteringly,

probably not. But in some characteristic way for that focal length and face. By distortion I mean

distortions to the form of the face, not color, contrast, bokeh, etc. (though there will be likely variations

with those factors as well).

 

Now if I take the same face and fill the frame of a DSLR with the 35mm lens how will the face look

relative to the first picture?

 

Unfortunately I don't have the right combination of lenses to perform the tests that would answer my

question.

 

But I do suspect the two "looks" will be different. Which is better or whether the difference matters is

probably a function of the face in the photo and the taste of the viewer.

 

If there are differences in the looks then while the fields of view might be the same the picture-taking-

potential is different.

 

Has anyone done this test?

[Hector Javkin]

If you shoot from the same distance you will have the same "distortion." This means that, if you use a 50mm lens on a 35mm camera, and an equivalent focal length on a 1.5X digital, and fill the frame with the same face on both, you will have the same effect. Of course, a 35mm lens is a tiny bit longer than the equivalent (33.3mm would be closer), so the apparent distortion would reflect that, but it is so close that you would hardly notice.

 

The only reason that certain focal lengths are often favored for portraits is that, for a given framing, they make you stand at a distance that gives a pleasing perspective.

[Ed_Ingold]

The "distortion" to which you refer is due to perspective (or point of view), which is solely dependent on distance, and not at all on the focal length. At a given distance, the focal length only affects the angular field of view, which is smaller with a sensor smaller than film.

 

If you shoot an head shot with a 50mm lens on a film camera, the facial features closest to the lens become exaggerated, due to perspective. If you stand at the same distance with a DSLR with the same lens, you will not get the full head view, but the distortion in what you do get is exactly the same. Now, if you move back until you have the same head shot composition as with film, you will be at the same distance as if using a 75mm (medium telephoto) lens with film. The distortion at this distance will be much less - most say more natural.

[Ed_Ingold]

A 35mm lens on a DSLR has about the same angular field of view as a 50mm lens on film. If you compose an head shot with a film camera and a 50mm lens, then take the same shot with a 35mm lens and a DSLR, the "distortion" or perspective would be identical. The exact field of view would be slightly different, because 35mm x 1.5 equals 52.5 mm (equivalent).

[nee_sung]

What has been said are all valid.

 

However, if you use a 35mm lens originally designed for 135 film, then that lens will tend to have more distortion built-in. That is because the lens will have to accommodate an image circle sufficient to cover the 135 film area. To correct that lens to the same standard of distortion as a 50mm lens would be expensive.

[jcuknz]

Which is why Bob suggested the use of a 50mm lens on a x1.6 crop factor DSLR for portraiture on a parrallel thread. You have to move back for equivalent framing and so you get a better perspective.

 

Even better would be to use an 85mm lens to equate with the 135 many use[d] for portraiture if their studio is big enough :-) I think Leitz supplied an 85mm once upon a time.

[ronaldo_r]

"...But I do suspect the two "looks" will be different..."

 

Michael - that's exactly what I thought some time ago, however this logic is incorrect.

 

PS: Apparently even fish eye perspective is normalised if you get close enough...

[mark u]

Read this:

 

http://www.pinnipedia.org/optics/distortion.html

[michael_dakin]

<p>

Thanks for the informative replies and reference! "Perspective distortion" is exactly the

factor I was concerned about. I'm sure the point about the 50mm being engineered more

perfectly than the 35mm lens is true but I suspect those differences esp. in the center of

the frame, would not really be problematic for most faces.

</p>

<p>

Armed with the correct terminology (thanks to the replies) I found another nice reference

concerning this subject, "<a href="http://heim.ifi.uio.no/~gisle/photo/crop.html">Effects

of cropping</a>" by Gisle Hannemyr. A couple of enlightening example photos are

included.

</p>

<p>

I've still not been able to find any side-by-side example photos involving faces. But I'm

not worried given that all the theory seems to be consistent. And my own photos taken

with the 50mm lens/1.5xDSLR of faces looked strangely good (which is what got me

thinking about this subject in the first place). Seeing evidence is always most comforting.

</p>

<p>

Thanks to everyone!

</p>