View Full Version : Normal and Expected? or Problem Lens
canoli
08-21-2009, 07:34 PM
Just wondering what you guys think - to me the distortion in this image seems awfully severe for a Canon L - it's the 17-40 at 17mm on a 40D.
I know all wide angle lenses distort, particularly zooms, but this seems a little overboard to me.
/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.29.05/pincush_5F00_web.jpg
canoli
08-21-2009, 08:01 PM
Here's another one - maybe it'll help to see 2 shots - same camera/lens, 40D, 17-40L at 17mm.
Is there a secret to minimizing this effect, or any tips to hide it?
Thanks for taking a look - and whatever thoughts you can share.
/resized-image.ashx/__size/800x533/__key/CommunityServer.Components.UserFiles/00.00.00.29.05/pincush2_5F00_web.jpg
clemmb
08-21-2009, 10:13 PM
I think expected. I have no experience with this lense but I notice this with my 24-105L. My 24-105@24 has 1.8% distortion. The 17-40@17 has 2.47%. The 16-35@16 has 2.37%. Check out Lens Tests / Reviews at http://www.photozone.de/canon-eos
Mark
piiooo
08-21-2009, 10:18 PM
This is a case of perspective distorsion. It'smore visiblewithwide angle lenses. My 17-40 L does the same.
It can be corrected by holding your camera level, you may have to elevate the camera to achieve a similar crop. Another way of correcting it would be a tilt and shift lens. You can also take care of it in post.
Perspective distorsions are very cool when used creatively.
Don Burkett
08-21-2009, 10:50 PM
This is a case of perspective distorsion. It'smore visiblewithwide angle lenses. My 17-40 L does the same.
It can be corrected by holding your camera level, you may have to elevate the camera to achieve a similar crop. Another way of correcting it would be a tilt and shift lens. You can also take care of it in post.
Perspective distorsions are very cool when used creatively.
<div style="clear: both;"]</div>
So does mine and a level camera is great advice. They are also very easy to clean up in post. PSP has a specific tool for this. Just remember the perspective correction will result in a cropped shot so leave room around the outside edges that you expect to loose. The closer the object the greater the influence the disortion has on what you see.
piiooo
08-21-2009, 10:58 PM
I think expected. I have no experience with this lense but I notice this with my 24-105L. My 24-105@24 has 1.8% distortion. The 17-40@17 has 2.47%. The 16-35@16 has 2.37%. Check out Lens Tests / Reviews at http://www.photozone.de/canon-eos
Mark
I think they discuss barrel distorsion in the photozone review. This phenomenon also shows up at wide angle, and is actually what fisheye lenses are about.
Perspective distorsion on the other hand is a result ofthe sensor planenot being parallel to the object plane.
canoli
08-22-2009, 12:13 AM
Thanks for passing along the website address clem - seems like a good site...
Part of their "verdict" on the 17-40 (on APS-C sensors) says:
"On the downside the level of distortions could haven been a little better at the wide-angle setting"
I may not use 17mm anymore - this one is at 23 and it's a lot better I think.
/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.29.05/_5F00_MG_5F00_0683_5F00_23mm.jpg
canoli
08-22-2009, 12:17 AM
Thanks for the comments guys - hadn't read them before I posted this last one.
I'm sure the camera was all over the place when I shot these - about a 40mph wind blowing up there, the lightning was flashing, thunder booming - pretty exciting storm - the blaring rain started about 30 secs after my last shot. Fun stuff!
canoli
08-22-2009, 12:23 AM
Is the 10-22mm similar to these - the first and second ones? I messed around with the 10-22 for a little while at B&H but I ended up liking the feel of the 17-40 so I got instead...
'course there's no harm in having both I guess! Do you guys get less distortion with the 10-22 or is it basically the same?
wickerprints
08-22-2009, 02:26 AM
I'm not sure if you're understanding what some people mean by "distortion."
Usually when photographers speak of lens distortions, they mean geometric distortions of a nature that deviates from a rectilinear projection--that is to say, certain features that form straight lines in the subject are imaged as curved lines. The two most common terms to refer to this type of distortion is "barrel" and "pincushion," the former describing a curvature outward, and the latter describing a curvature inward.
However, based on the images you have posted, what you appear to be describing as "distortion" is actually the natural consequence of a rectilinear projection. It is called "perspective distortion" but it is the inescapable result of trying to project a curved field of view onto a flat plane. The shapes of the buildings taper upward not because the lens has high distortion, but because it is a geometric necessity to keep the straight lines of the subject straight on in the image. The same effect is observed if you stand next to a tall building and look up the side--your eyes will perceive the sides of the building as two lines converging near the top, even though in reality they are parallel.
Distant objects appear smaller than closer objects. This is a critical property that the human brain uses to infer the distance of an object.
The wideness of the 17-40/4L lens is such that this rectilinear perspective is very pronounced. Our eyes do not have such a short focal length and therefore we are unaccustomed to this extreme wide-angle perspective. The tapering of the buildings is because you have tilted the camera slightly upward--just like the skyscraper example I mentioned.
The barrel distortion @ 17mm is there and it is noticeable, but it is very subtle in the images you posted. It is easily corrected in Canon's DPP software. It is not a lens defect--it is part of the compromise of designing a wide-angle zoom lens. Many lenses, including the 24-70/2.8L, 24-105/4L IS, 16-35/2.8L II, and yes, even the EF-S 10-22, exhibit barrel distortion at the wide end and pincushion at the long end. Primes show less distortion but it too is present. But I again emphasize that the barrel/pincushion distortion we're talking about here is NOT what you are seeing in the attached images.
If you purchased the 10-22, you would find your "distortion" (which is really a property of projecting a curved field of view onto a flat plane) to be even more dramatic, as the 10mm focal length is dramatically wider than 17mm.
canoli
08-22-2009, 11:58 AM
Oh, okay, I see what you're saying - I've got a bad lens here, that's all there is to it. I should go to B&H and demand my money back, or send it off to Canon for recalibration. Like you said, if I'd bought the 10-22 I wouldn't have to deal with this problem.
(!)
No. But seriously wick - thanks for the explanation. I thought I "knew" about the effect you describe, but I haven't seen it so pronounced until yesterday. Everything I've shot with the 17-40 up till now has been true landscapes, focused at infinity and without any buildings nearby. I'm guessing that's why I haven't noticed much of this "distortion" before.
So perspective distortion will appear stronger the more you raise the lens (shoot upward). To minimize the effect (besides using a tilt/shift) you can keep the camera level. Is there anything else?
Thanks again wicker - and thanks everybody for the comments, I really appreciate it.
peety3
08-24-2009, 04:16 PM
So perspective distortion will appear stronger the more you raise the lens (shoot upward). To minimize the effect (besides using a tilt/shift) you can keep the camera level. Is there anything else?
Keep the camera level. I was in Philly two months ago, and my girlfriend was shooting City Hall. I told her to aim at a point about 4'9" above the ground (rough estimate of her eye level). "But we'll have lots of street, and that's not what I want!" "I know dear, but we can crop that out." We ended up walking back a block or two so the top half of the frame could capture the entire building.
And don't forget to LOOK at what you're shooting, from your shooting position. A tilt/shift lens can correct the perspective distortion, but you'll still have the view from that point. If there's a balcony and a cool door leading to the balcony, you won't see the door at all shooting from too close to the building.
Chuck Lee
08-24-2009, 06:02 PM
canoli,
Not only hold the camera level to the horizon but purchase PTLens ("http://epaperpress.com/ptlens/). It's a fantastic little program or PS plugin.
Wait till you see the perspective distortion with that lens on a full frame. Or when you shoot with a 10-22 @10mm on APS-C, thoughthat lensappears to bevery rectilinear based on one of the examples Sean S. was showing. Yeah, photozone.deillustrates very wellthe barrel and pincushion distortion for most commom wide angle lenses.
I'm not sure if you're understanding what some people mean by "distortion."
Usually when photographers speak of lens distortions, they mean geometric distortions of a nature that deviates from a rectilinear projection--that is to say, certain features that form straight lines in the subject are imaged as curved lines. The two most common terms to refer to this type of distortion is "barrel" and "pincushion," the former describing a curvature outward, and the latter describing a curvature inward.
However, based on the images you have posted, what you appear to be describing as "distortion" is actually the natural consequence of a rectilinear projection. It is called "perspective distortion" but it is the inescapable result of trying to project a curved field of view onto a flat plane. The shapes of the buildings taper upward not because the lens has high distortion, but because it is a geometric necessity to keep the straight lines of the subject straight on in the image. The same effect is observed if you stand next to a tall building and look up the side--your eyes will perceive the sides of the building as two lines converging near the top, even though in reality they are parallel.
Distant objects appear smaller than closer objects. This is a critical property that the human brain uses to infer the distance of an object.
The wideness of the 17-40/4L lens is such that this rectilinear perspective is very pronounced. Our eyes do not have such a short focal length and therefore we are unaccustomed to this extreme wide-angle perspective. The tapering of the buildings is because you have tilted the camera slightly upward--just like the skyscraper example I mentioned.
The barrel distortion @ 17mm is there and it is noticeable, but it is very subtle in the images you posted. It is easily corrected in Canon's DPP software. It is not a lens defect--it is part of the compromise of designing a wide-angle zoom lens. Many lenses, including the 24-70/2.8L, 24-105/4L IS, 16-35/2.8L II, and yes, even the EF-S 10-22, exhibit barrel distortion at the wide end and pincushion at the long end. Primes show less distortion but it too is present. But I again emphasize that the barrel/pincushion distortion we're talking about here is NOT what you are seeing in the attached images.
If you purchased the 10-22, you would find your "distortion" (which is really a property of projecting a curved field of view onto a flat plane) to be even more dramatic, as the 10mm focal length is dramatically wider than 17mm.
<div style="clear: both;"]</div>
Wickerprints, thanks for this explanation, it is absolutely fantastic[Y] I actually purchased 10-22mm recently and being a new in photography had a bit of an adjustment issue with it, I was shooting duds for a while until I figured the whole tilt problems, or in other words "operator error" [:$] It took me a while to figure it out what you put so eloquently and easy in your explanation. Thanks again. You guys are the best!