can't wrap my mind around these distortions

Hello,

I'm seeing some strange lens distortion issues with the Sigma 18-50 2.8. I'd love to get some feedback on this, because this just does not fit into my head right now. The following images were all taken with the Sigma 18-50 2.8 at the widest end, 18mm, recently on the beautiful San Francisco bayfront (exif is probably not there, since these are processed in photoshop cs). The first image has some obvious pincussion (!!) distortion, that can be (somewhat) corrected in the following (second) image. Normally the Sigma 18-50 has some barrel distortion at the wide end as usual, so isn't this quite weird? The next (third) image is taken with the same lens soon after, and has some barrel distortion now, which can also be (somewhat) corrected (fourth image).

Am I completely misinterpreting what I see? How can I have different types of distortion with the same lens at the same focal length?

There was a circular polarizer on the lens, but I cannot see how that would affect the distortions of the lens. Any help?









The pics are hosted on my computer at home, so they may disappear after a while..

--
Beauty is in the eye of the photographer

Can't see the shots, but if you use PT Lens and have the wrong lens selected it can overcorrect barrel distortion. That's the only thing that immediately comes to mind without seeing the pics in question.

Let's try again with the images. Apparently my ISP blocks the default http port 80, so I had to work around to a different port (when I preview, the links work fine since I'm accessing locally).

PZ, I agree with your PTLens trick, - I actually searched around for another plugin that gives you manual control over the correction amount, and that's how I corrected these images (Richard Rosenman lens distortion corrector). But I'm just surprised that the default PTLens correction for this lens does next to nothing. I have no reason to doubt the PTLens parameters for this lens are off.. I just want to understand what's going on - how can I have so much pincushion at 18mm?

Sorry for the size of the images - please use the Zoom out thing a few times to bring it to a decent size.

images are too big to see side by side....maybe crop sample areas and put side by side, if an element is not seated correctly in the lens barrel , it may be moving around and thus giving you the variety of distortions you are encountering. One internal element can shift, that is designed to correct distortion, and may enhance distortion, fun, eh?
--
thank Canon for L and BMW for M

wow, this would be something unusual of an explanation, but it does make sense. sounds like a serious problem, I guess I'll test the lens shooting some squares or something. Thanks, Ted

You can try and run them through either PTLens or DXO Optics. (DXO has a trial period where it is fully functional.) If DXO can compensate for it then your lens is probably ok. (That is assuming it supports your camera and lens configuration.)

fedor said:

wow, this would be something unusual of an explanation, but it does
make sense. sounds like a serious problem, I guess I'll test the
lens shooting some squares or something. Thanks, Ted

Click to expand...

--
Peter Chung

You don't have pincussion distortion in the first photo, you've got the effect of perspective. Perpective is very visible in a photo such as this; you are using a wide angle lens with the lens pointed upwards a bit. This will cause verticals to appear to converge towards the top of the frame. The opposite will occur if you were pointing the camera downwards; the verticals would appear to converge towards the bottom of the frame. Keep the camera level and you won't see this.

Here's a downsized copy of your first photo with the perspective "corrected" in photoshop:



--
Galleries: http://www.pbase.com/jon_b

Your sensor has to be parallel to the buildings (vertical) for them to show up as having parallel sides. Otherwise they're like railroad tracks and will appear to converge. If you point your camera straight down at railroad tracks (sensor parallel to the tracks), they will be parallel in the picture. But if you put the camera even a tiny bit towards the horizon, the tracks will not appear parallel.

But it is possible for a lens to have different distortions at the same focal length, if the focus distance is different. That's not the case here, but it is possible.

Thanks jgb and John Kim! I hadn't appreciated that the tilt of the camera relative to parallel lines in the picture makes a difference (and a lot of difference, as I now see). I guess these images will require more PS than I anticipated

fedor said:

Thanks jgb and John Kim! I hadn't appreciated that the tilt of the
camera relative to parallel lines in the picture makes a difference
(and a lot of difference, as I now see). I guess these images will
require more PS than I anticipated

Click to expand...

Or, you could get a tilt-shift lens.

--
Lee Jay
(see profile for equipment)