Distortion at the Edge of Wide Angle Zooms

In a current magazine there are several articles illustrated by the same photographer. After looking at the photo credit for the first article, I was able to pick out all of the shots that he did throughout the rest of the magazine by looking for distorted heads in his pictures. It looks like he shot these all with a wide angle lens and had the subjects near the frame edges.

I've not had this trouble using my wide zooms (17-35 and 16-35) for event coverage, even when people are close to the edges. However, the 10D is cropping the edges of these shots quite a bit, so in effect people at the edges of my photos are not near the edges of the images created by the lens.

As we move to 1.3 crop factor and full frame cameras, I wonder if this is going to be more of a problem with the aforementioned lenses. My EXIF data shows that I often LIVE near the 17 mm setting. I can always set the lens to 21 mm (to get the equivalent angle of view on a Mark II), but will this solve the problem? I think that it might.

Has anyone who has made the transition to a 1D or is shooting full frame cameras with these lenses experienced this effect?

Maybe I'm just rationalizing the purchase of a 24-70/2.8L ...

Thanks.

Mike

Distortion at the edge of Wide angle zooms

Hello:

The 16-35, and moreso, the 14mm will appear to distort more as you get closer to the edge of the frame. With our D60, the effect is less noticeable because of the 1.6 chip factor. With my full-frame EOS3, and the full frame 1DS it is much more noticeable. The 1.3 MKII will be somewhat less noticeable @ the edges. I don't find this effect to be a problem. You can position subjects to minimize the effect. If you can't control a subject's position, I would say live with it. It's not an unusual effect. I see it all the time in publications.

On a somewhat unrelated note, there is also more color fringing with the digicams at the edge of the frame with the 16-35 @16mm. Even the 14mmL lens exhibits this problem. PSCS has a setting in the raw converter that minimizes this problem, but yo need to utilize split exposures with the camera on a tripod, and blend the two exposures.

Tanks!

That's what I suspected.

I guess the mitigating thing is that shots at events (weddings, fund raisers, etc.) are really reportage, rather than portraits, which are usually intended to flatter the subject. For portraits, I'm much more likely to use a longer lens that doesn't have that problem. The only exception that I can think of would be an environmental portrait, and for those, I would expect to have enough control of the situation to keep the subject away from the edges.

Thanks for the helpful reply.

Mike

You may want to check out DxO Optics Pro, a soon to be released post-processing software for correcting image distortion. It's pricey ($100+) but apparently quite good. Michael Reichmann recently wrote up a very postive review.

Did you realize that it's physically impossible to completely "correct" pictures? The reason for this is that you squeeze three dimensions into two. By doing this, lenses are normally corrected for rectilinearity (all straight lines/edges appear straight). The disadvantage of this correction is that spherical 3D objects appear egg-shaped. When you correct these pictures for round-shaped spheres, you end up with a picture with a fisheye look. As long as you don't have straight lines/edges in the frame, these pictures look more "natural" than the rectilinear ones. But as soon as you have straight lines which don't go through the center of the frame, they become bent and that looks "wrong" to our eyes (although it isn't).
One can't "correct" for both, the rectilinearity and 3D objects. Not even with "L"

--Jens--

Mike H wrote:
In a current magazine there are several articles illustrated by the same photographer. After looking at the photo credit for the first article, I was able to pick out all of the shots that he did throughout the rest of the magazine by looking for distorted heads in his pictures. It looks like he shot these all with a wide angle lens and had the subjects near the frame edges.

I've not had this trouble using my wide zooms (17-35 and 16-35) for event coverage, even when people are close to the edges. However, the 10D is cropping the edges of these shots quite a bit, so in effect people at the edges of my photos are not near the edges of the images created by the lens.

As we move to 1.3 crop factor and full frame cameras, I wonder if this is going to be more of a problem with the aforementioned lenses. My EXIF data shows that I often LIVE near the 17 mm setting. I can always set the lens to 21 mm (to get the equivalent angle of view on a Mark II), but will this solve the problem? I think that it might.

Has anyone who has made the transition to a 1D or is shooting full frame cameras with these lenses experienced this effect?

Maybe I'm just rationalizing the purchase of a 24-70/2.8L ...

Thanks.

Mike

I used a Nikkor 20mm lens for years with Nikon film cameras. It's about what a 12.5 mm lens would be on my 10D. There is absolutely distortion if you put something round - like a head - in the corners. So you don't.

A head anywhere in the center of the frame will have no distortion at all and you work with that. I've done magazine shots where I've used the 20 and ignored the rule of thirds. This was a vertical portrait shot, where I was trying to use some journalisitic approach. Normally the subject's eyes would be 1/3 of the way down from the top of the frame. In this case I put the eyes lower. If you looked at the raw transparency it would look like I ignored rule of thirds, and didn't fill the frame particularly well. But I fully intended to crop the top out of the image. I preserved the no-distortion face and still got a super-wide angle interpretation in the bottom of the frame.

EXA1a wrote:
Did you realize that it's physically impossible to completely "correct" pictures? The reason for this is that you squeeze three dimensions into two. By doing this, lenses are normally corrected for rectilinearity (all straight lines/edges appear straight). The disadvantage of this correction is that spherical 3D objects appear egg-shaped. When you correct these pictures for round-shaped spheres, you end up with a picture with a fisheye look. As long as you don't have straight lines/edges in the frame, these pictures look more "natural" than the rectilinear ones. But as soon as you have straight lines which don't go through the center of the frame, they become bent and that looks "wrong" to our eyes (although it isn't).
One can't "correct" for both, the rectilinearity and 3D objects. Not even with "L"

--Jens--

Agreed. Image distortions related to perspective and geometry are apparent, not real. But unless the viewer of an image has the same viewpoint as the camera (i.e. the same camera to subject distance and perspective), the human visual apparatus will perceive distortion in the image. With ultra wide-angle, this is especially apparent since human vision cannot actually duplicate this perspective. Thus we can make "distortion corrections" to please the mind, the same reason we buy "L" lenses.

Thanks for the additional responses, they were very helpful. For now, the plan will be to keep using my 16-35 as a walk-round lens when the Mark II arrives, and watch the corners.

If it turns out that I want something a little longer for my standard lens, I'm going to go with the 24-85/3.5-4.5. That lens is MUCH smaller, lighter, and cheaper than the 24-70/2.8L, and its MTF numbers on Photodo.com look good (even if a little below L standards). The lighter weight and price will help make up for those characteristics of the Mark II.

Mike