Is Barrel Distortion caused by user error or flawed gear?

[naga]

Well I finally bought the Nikon D7000 and Sigma 17-70mm f/2.8-4 DC Macro OS HSM Lens for Nikon Mount Digital SLR Cameras. All my photos suffer from barrel distortion. I've been doing my best to read the manual but it's progressing fairly slow with the limited free time I have. Can you point in my the right direction if this is a user error or flawed gear?

[kpmedia]

Short answer:

Q: Is Barrel Distortion caused by user error or flawed gear?
A: Both.

Longer Answer:

There's going to be natural barrel distortion at wide angles. There's also going to be smearing and softness on the corners as you approach ultra-wide angles (10-16mm). It's an unavoidable artifact of that angle of view. It's closer to how a fish sees, than how a human sees.

The more flat you are (planar) to the object, the less distortion. The more you angle away from a flat plane, the more obvious distortion gets. For example:

You | | Subject - same plane
You | / Subject - angled off the plane, more distortion
You / | Subject - angled off the plane, more distortion

Distance makes a difference, too. You can easily give somebody a fat head and a little body. A few years back, I took some photos of a dog, where his head was twice the size of the rest of his body. I was both close up with a wide angle (~14mm), and at an angle that made the effect more intense.

How does it shoot at 50mm and 70mm?

I'll send you an email, too. I can bring you a lens (or two or three) of mine, if needed, and we can compare some higher-end Tokina and Nikkor glass against that Sigma lens, on the same body.

[jmac698]

This can be corrected in software.

Some tools let you play with sliders until it looks right. For best quality and convenience, you need a camera profile. If the program doesn't have a profile for your lens, you can make one. Creating a profile usually involves printing out a calibration target (like a checkerboard pattern), and taking pictures of it at different angles.

Photoshop tutorial:
http://www.kenrockwell.com/tech/corr...distortion.htm

PTLens:
http://epaperpress.com/ptlens/

Gimp (free Photoshop clone)
http://www.gimp.org

Dxo
http://www.dxo.com/us/photo/dxo_optics_pro/features

For the serious photographer, there is a lot of postprocessing done. You would convert from RAW files, removing chromatic abberation, lens distortions, vignetting, and then color adjustments, sharpening, cropping, noise reduction, even HDR.

Using the RAW files allows you to recover detail in washed out hgihlights or dark areas.

I will mention http://rawtherapee.com/blog/features for free RAW conversion.

Other fun things you can do include focus stacking, HDR, high-speed photography (with fast flash), low-light photography, using polarization filters, extracting 3d models from photos, removing hot pixels, dark frame subtraction, time-lapse, focus correction, panoramas, architectural views, stitching (for large things like mapping or objects), perspective correction, and all sorts of astrophotography techniques.

If you had a Canon you could use http://chdk.wikia.com/wiki/CHDK which is a custom firmware to add new features.

[kpmedia]

Very true! Adobe Photoshop CS3/CS4/CS5/CS6 and Adobe Lightroom 3 have lens correction filters.

The best filters are from DxO, and those are based on hardware measurements. You have to build profiles from the body and lens database. The problem, however, is that it's almost purely high-end bodies and lenses. It has very few third-party manufacturers in the db, and is almost all Nikkor and Canon L glass. Because I have a mix of maker glass, and third-party glass, I don't much care for DxO.

[jmac698]

You can use the free Adobe Lens Profile Creator
http://labs.adobe.com/technologies/lensprofile_creator/

Install it and do a search within the program, users may have submitted some of your lenses.

There's many free ways to do this, but they get more cumbersome. I don't think Adobe's procedure is the best algorithm, but I'm sure you don't want to run Matlab to do this

You can do your own hardware measurements, and while you may not have a large accurately printed target and special lighting, the algorithms exist to find distortion perfectly well, to several decimal places accuracy. It's just math, but the more pictures you take, the better. Vignetting is a different story, but it happens that the natural blue sky can make a pretty even light source.