12-24 DX zoom distortion worse than 18-36mm in full-frame lenses

[roger_s]

Before I bought my 12-24 Nikon DX zoom for DSLR, I searched the internet for

opinions, and almost all praised the lens. There was one blog opinion,

however, made by a working-photojournalist who said the distortion of the 12-

24 was more severe compared to equivalent focal lengths in full-frame lenses,

in particular with people photos.

 

Since the weight of positive reviews outweighted that one negative blog entry,

I felt confident to purchase the 12-24 DX lens.

 

However, after useing the 12-24 lens for many months, I am growing uneasy

about the amount of distortion I see when using the 12-24 for people pictures,

when the subject is less than 2 meters away. Sometimes a person appearing at

the edge of the photo, standing upright, can appear to be tilted 45 degrees.

 

Now, I fully understand that a degree of distortion is inherent for superwides

such as a 12mmDX (18mmFF), however, I wonder if there are any people who have

used the 12-24, AND who have also used the equivalent focal lenghts in Full-

fame 35mm, who can compare the level of distortion of the 12-24 with their

equivalent FF lenses.

 

It is just that, in magazines, I see a lot of superwide photographs of close-

in people photos, that do not seem to display the extent of distortion I see

in photos made by the 12-24DX.

 

Note that this problem is not pronounced when the subject is distant. e.g.

there is no problem using the 12-24 @ 12mm for long distance landscapes.

 

The problem I am referring to occurs when taking people photos where the

subject is less than 2 meters away, and appears close to the right or left

edges of the photo.

 

I don't need comments from people who've never used the 12-24, saying that

distortion is inherent with close-up photos with ultrawides. That is obvious,

so please don't reply with those type of comments. I am specifically wanting

comments from people who have used the 12-24DX, and also have prior experience

with equivalent focal-lenghs in FF lenses.

[jbq]

How about a sample?

[sansmeat]

Capture NX has improved distortion control to help with this problem. Adobe's raw convertor has tools that will help as well.

 

~b~

[ShunCheung]

Roger, you indeed need to post an example. When we discuss lens distortion, we are talking about straight lines, especially those near the edge of the frame, become a curve or even a weavy line.

 

You seem to be talking about converging parallels similar to the effect of shooting up a tall building from the ground level with a wide lens. That is not distortion.

 

Attached is my standard gargae door shot with the Nikkor 12-24mm/f4 at 12mm. There is some complex distortion but nothing too unusual for such a super-wide lens.<div></div>

[roger_s]

The EXIF data indicates that both these photos were taken at 12mm on the 12-24.

[roger_s]

Here's another one. In both photos, note the faces of the boys at the edge of the photos.

[pcm__]

Hi Roger. I own an 18mm MF lens for my Nikon fm3, as well as the 12-24 for digital. The

distortion at 12mm on digital is much more pronounced IMO than the 18mm on full

frame.

 

Unfortunately, this is a side effect of smaller sensors that cannot be avoided with current

lens technology alone. a 1.5 crop does not affect mid to long lens shooters as much, but

us wide-angle fans suffer from side effects like the one you've mentioned.

 

As Nikon users, we have to live with it or go with the Canon 5d. I decided to live with it,

but I'm always careful these days not to catch anyone in the edge of the frame.

 

Best

[ShunCheung]

Roger, in your image, the way the face of the person on the right side is distorted is the standard effect of using a wide angle from a close distance. The building in the background look tilted because that building is not parallel to the plane of the sensor/film.

 

Those effects are all very normal for wide angles and are not lens distortion that is discussed as a defect or poor lens design. Any 12mm lens on a Nikon DSLR or any 18mm lens on a 35mm film SLR will give you the same effects.

[roger_s]

Shun, what do you think of PCM's comment?

[narayan_hegde]

I have been wondering the same. Have not seen comparison at aps focal length vs comparable FF focal lengths to see differences. But check out the DXO tool -

http://www.dxo.com/intl/photo/dxo_optics_pro/overview/flexible_geometry/anamorphosis

Which talks about the various distortions in wide angle lenses. I think the distortion you are seeing is -Spherical volume anamorphosis. DXO claims it can correct these distortions.

[ShunCheung]

Roger, I can tell you one more time, any 12mm lens on a Nikon DSLR or any 18mm lens on a "full-frame" SLR such as 35mm film SLR will give you the same distorted effect. It has nothing to do with the DX sensor size.

 

It is very easy to verify my comment if you have access to a 18mm lens on a film SLR. You can simply look through the viewfinders with 18mm on film SLR compared to 12mm on DSLR. If you are still not convinced, shoot some pictures of the same scene for an A/B comparison. If I had a FF DSLR, I would have done that and post the results here, but I am not going to go through the trouble to shoot some film, get it developed and scanned just to demonstrate it.

[Ed_Ingold]

I agree with Shun's assessment. The elongation of heads toward the edges and corners of a wide-angle shot is characteristic of a rectilinear lens. It is an effect of perspective and not distortion per se. Notice that it goes away if you view the image from a distance equal to the focal length times the magnification. It is present with any lens at any distance, but only becomes obvious when the angle of view is large.

 

The 12-24 does distort straight lines near the edges of the frame, including a "moustache" shape at some focal lengths. You might notice this when shooting buildings, but not for the typical street or landscape photo.

[jbq]

Roger, what you're seeing is the normal behavior of a rectilinear ultra-wide, where the geometry of the result feels un-natural. You'll probably get more satisfying results if you point the lens horizontally, so that verticals stay parallel (in both your examples the lens is pointed very sharply down).

[roger_s]

I'm presently downloading the large 33MB demo of DxO, however, I note another distortion correction software http://epaperpress.com/ptlens/index.html

 

I previously emailed the guy at ptlens, and he said his software would not cure distorted faces at edges, since he said, "PTLens will correct pincushion and barrel distortion, but does not correct projection distortion". However, the DxO website seems to promise correction of distorted faces at the edge of the frame.

 

I wonder whether Dxo corrects for rectilinear distortion caused by pointing the lens downward.

[dstate1]

Roger, you can correct for perspective distortion in PS or Elements. It is not the same

as correcting for barrel or pincushion distortion. The distortion in your shots, as

mentioned above, is due to the natural effects of an extreme wide angle lens used up

close. No wide angle is immune to this effect.

 

best wishes

Dan

[walterh]

Roger when you correct for the rectilinear imaging of the lens you will loose a large amount of your frame. It may be a solution for some shots - presumably not the one you showed (cutting the head off). You could use a fisheye lens and have fun with a different type of "distortion".

[stevea]

Here's a shot of my garage door taken from about 2 meters using a Nikon 12-24mm lens @12mm. Is this the normal perspective effect or is it distortion?

[ShunCheung]

Steven, you have the common "converging parellels" effect. See my garage door shot earlier. I made sure that my camera wasn't tilted so that the door was parallel to the film/sensor plane. Your image is similar to those shooting a tall building from the ground level while tilting the camera upward. We correct that problem with shift lenses so that the film/sensor plane is parallel to the building again or nowadays we can fix that in PhotoShop also.

 

BTW, please limit the size of your attached images. Anything wider than 800 pixels across is hard to read on a browser. We also suggest keeping the film size to within 100K bytes.

[ShunCheung]

Attached are a couple of images of the Santana Row shopping district in downtown San Jose in northern California. I shot it with the 12-24 DX at 15mm. The problems with converging verticles is very obvious. The lower one is after corrections in PhotoShop.<div></div>

[andy a.]

Just to be clear, what you are seeing is not distortion. It's a fact of perspective, and the wider the lens, the more pronounced it is. Since they have the same perspective, a 12 mm on dx will be identical to 18 mm on 35mm.

 

Further, this is not a "flaw" in the lens either. If you could see a sphere at the edge of your vision, it would appear elongated the same way through your eye, as strange as this seems.

 

Finally, software tools don't really "correct" this since it's not distortion. The photoshop perspective correction tool applies a distortion to an image to get a result more pleasing to the eye, e.g. vertical lines that are parallel.

[dennis lee]

Well Roger, it looks like only one person responded with an actual comparison, interesting

question. I've been thinking about the 12-24 myself. I too would like to see a comparison

although I must say that as a long, long time user of a 20mm on film (and the old 20/4 at

that) I don't find the distortion too severe. Actually, they look better than I would expect for

an equivalent of 18mm. Nice shots by the way. Let's hope we see a few examples.

[kaiyen]

First off, it seems like what you're asking for comparison is the 12mm end on the 12-24, and the 18mm end on, say, the 18-35. If you're talking about both on a DSLR, then obviously you're talking completely different focal lengths and FOV. If you mean the 18mm on a film SLR vs 12mm on a DSLR, then yes, there will be less corner distortion as the lens simply isn't as wide.

 

I ran into this problem at a wedding, and it's just a character of the lens. Of any lens of such a side angle. I needed the 12-24 because I had to get in close. So I just got in even closer, left a lot of space on the edges, and cropped them out. It's the difference between getting the shot or not in that situation.

 

Shun - I'd love to go shooting with you sometime, if you don't mind company. I keep seeing sample photos from you of places near me (I life 5 minutes from Santana Row) and you mention K&S, so we're clearly quite close. As someone so knowledgeable about Nikon's lineup DSLR and SLR, lenses, etc, I'm sure I'd find it quite educational. If you're game, of course.

 

allan

[ShunCheung]

Allan, I too live in the San Francisco Bay Area and took that image in Santana Row just last Saturday. I used it as an example since it was handy. For getting together, we can discuss in private e-mail.

[jake_mecklenborg]

Yes, this lens CAN produce good results, but its behavior is too erratic. In any kind of

situation where things are happening with any kind of randomness, having a lens where

you're afraid to use it at either its widest aperture or widest focal length (or worst of all,

both) is a major distraction. In the 6 or so months I had this lens, I don't think I actually

went with a full-frame 12mm shot more than 10 times. I did shoot at 12mm quite a bit,

but those images were usually cropped, and sometimes the distortion was just plain

bizarre. For architecture, again, this lens is very tedious to use at the 12-14mm range

because you have to be so careful so that you don't end up having to spend an hour on

Photoshop with at-best mediocre results. When you first get it, it all seems so easy but as

you get to know this lens, you come to realize it has sweet spots and it has typically

useless results at 12mm and f/4. Photoshop and distortion correction can improve on

anything, but it can't make a fantastic image out of low-costrast, distorted input every

time. If you are shooting as a hobby or for a low-res newspaper, it's less of an issue than

shooting for magazines or making large prints. Knowing that you are going to have to

crop the image significantly to correct perspective while shooting helps, but you are losing

tons of resolution. Obviously this matters less for the web and low-res printing, and you'd

be using a view camera for pro architectural work anyway.

 

Also, as a photo-j lens, this lens is actually pretty good in my opinion for a good number

of situations because the AF is good and with f/4 as the max aperture, you have more in-

focus shots than with a faster lens. I photographed tons of bands in low-light situations

and got a lot of great shots with this lens. The dark surroundings mask the distortion

issues or even make that type of image more interesting.<div></div>