The effect of software distortion correction on lens sharpness

Started 7 months ago | Discussions thread
ForumParentFirstPreviousNextNext unread
Flat view
Dr_Jon
Senior MemberPosts: 1,107Gear list
Like?
The effect of software distortion correction on lens sharpness
7 months ago

(Note - this post is really just for people interested in the technical side of m43, it doesn't have anything to say about how good or otherwise the m43 system is. I like it BTW.)

I may regret this, as I really don't want to start an argument, but I found this very interesting so thought I'd share...

I've always suspected the distortion correction may effect the edge sharpness of some lenses a bit more than I'd like. (Other opinions are available!) Partly as some of the lenses with the most correction (e.g. my 12-35 at the wide end) have more of a drop in edge sharpness over centre sharpness than I'd like (but they might just not be that good of course) and partly as lens rentals who use uncorrected data seem to get better edge sharpness relative to centre sharpness than the others (photozone, ephotozine, lenstip) who use corrected data (alas nobody uses DXO Optics Pro corrections, which do even better than the native ones, e.g. for my 12-35 even after non-DXO "correction" there's a fair bit of distortion left ).

Anyway, I was trying to get photozone to do some tests with and without distortion correction figures and I have no idea if I had any part in it, but some interesting MTF graphs have appeared on their site for a couple of Samsung NX lenses...
http://www.photozone.de/samsungnx/696_samsungnx30f2?start=1
http://www.photozone.de/samsungnx/694_samsungnx20f28?start=1
(Mouse-over the MTF graph.)
Which while not being m43 lenses perhaps does give an idea of the magnitude of the effect.

Of course what it won't do is answer the question of whether the lens+correction gives a better result than a lens designed to work without correction. Again a variety of opinions are available, but little evidence so please don't jump up and down without some supporting data, this is just a little post with a couple of graphs. My personal feeling is visually the correction doesn't lose you much, but on the actual fine measurement of sharpness it loses more than you'd get with a good non-corrected lens (for a lens needing a reasonable amount of correction). I take this view from looking at image resizing algorithms and looking at what they do to fine detail (as the correction is basically just stretching/shrinking various parts of the image, using who knows what algorithm).

Please don't argue too much, like I said I just found the graphs interesting... I'm still happy with the sharpness of my better m43 lenses (20, 12-35, 35-100) and this is just out of interest.

John

 Dr_Jon's gear list:Dr_Jon's gear list
Fujifilm FinePix Real 3D W3 Sony RX100 Canon EOS 5D Mark II Panasonic Lumix DMC-GF1 Panasonic Lumix DMC-GH3 +23 more
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