Software distortion correction works - here's the proof

Started Jul 7, 2013 | Discussions thread
ForumParentFirstPreviousNextNext unread
Flat view
knickerhawk
Veteran MemberPosts: 3,442
Like?
Software distortion correction works - here's the proof
Jul 7, 2013

The thread started recently by Walkaround regarding the supposed inferiority of m4/3 lenses due to lens distortion corrected in software quickly went to the limit before I had a chance to respond to the following post:

http://www.dpreview.com/forums/post/51761806

The claims made in that post closely resemble the claims made by others in the thread regarding the purported inferiority of software-based distortion correction versus optical correction. Others responded that, actually, software correction is very effective and in some regards might actually be superior to optical distortion correction (there are no free lunches when it comes to adding more glass). Unfortunately, nobody in the thread offered anything in the way of empirical evidence to support their claims regarding the inferiority/superiority of software correction. Let's see if we can make a thread that deals with the question in an evidence-based way instead of based on speculation, guess work and hyperbole (in either direction). I'd like to kick it off with these samples. I'm using a test shot from my OM-D using the 12-50mm kit lens. The shot was at 25mm and processed from RAW in ACR with only mild sharpening and no other adjustments. I then applied a significant amount of barrel distortion and saved the image as a TIFF. The distorted TIFF version is the first shot below. The distortion scaling in ACR doesn't correspond to the percentage distortion usually cited in lens reviews so I can't tell you exactly how much distortion this is in terms of percentage, but as you can see it's quite significant and more than the 5% range you typically see for many of the m4/3 lenses.

Next, I took the distorted TIFF output file and tried my best in ACR to correct the distortion. Important: this "repair" work was done on a TIFF output file, not the original ORF or PSD file that "remembered" the original image, so it was not a trivial exercise of just reversing the settings I had previously made. It actually required manipulation of the distortion, vertical, scale and aspect sliders to get a reasonably close match to the original undistorted image. Posted below the full distorted image are 100% crops from the extreme lower middle edge and extreme upper right corner. These are areas where the amount of distortion is greatest, so if there is any adverse effect of software correction of distortion it should be evident in these areas. To my eyes, there is no meaningful visble degradation in the software corrected versions at 100%. At 200% I can see the tiniest (and I mean "tiny") difference that's easily removed by a very light application of added sharpening.

So, can you tell which crops are from the original image and which came from the significantly distorted version? If you were a non-believer in the powers of software correction and the need for optical correction, are you still so certain that optical is the only valid way to deal with the issue? If so, why and where's your proof?

ForumParentFirstPreviousNextNext unread
Flat view
Post (hide subjects)Posted by
ForumParentFirstPreviousNextNext unread
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark post MMy threads
Color scheme? Blue / Yellow