Correcting horizons: is less rotation always better?

mike3996

Legend
Location
Finland
Hi to all signal processing experts.

Sometimes my pic is off by 2.5 degrees; sometimes by -0.14. Intuitively one could imagine that the image degrades less if you have to perform less rotation.

Then again, rotation by a multiple of 90 is effectively lossless.

I currently assume that the image gets a constant amount of degradation from a rotation that isn't a multiple of 90. Of course I never could tell any degradation from a 24 megapixel image, but I'm intrigued if there is some theory to this. Mathematics can be insightful -- perhaps the picture comes out perfect if it's a bayer image and the rotation is something absurdly silly like e/360+pi
 
Can I ask why this matters? Are you saying you'll keep an image thats skewiff if it means it wont degrade at all? I'm failing to comprehend this, entirely.
Purely theoretical pondering, that's all :)

I don't notice any degradation in a rotated image but the math says it happens. There's also the aspect of psychovisuals, ie degradation may not even be a measurable property of an image.
 
Purely theoretical pondering, that's all :)

I don't notice any degradation in a rotated image but the math says it happens. There's also the aspect of psychovisuals, ie degradation may not even be a measurable property of an image.
I had a subscription to the Mirrorless section on diglloyd.com of Lloyd Chambers, one of the most avid pixel-peepers I have ever seen. He showed the degradation in resolution of photos by applying correction for geometrical distortion; I guess it's the same kind of degradation you get when rotating or doing transforms in Photoshop or Lightroom. Even when zooming in I'm in doubt if I was able to see that degradation, so I continued not to care. I must say I don't care about stuff on pixel-level ever since I got the Sony A7R2 with its 42 MP sensor and it's going to be less and less significant with the ever-increasing resolution of newer sensors.

BTW, I ended my subscription last May because I couldn't be bothered anymore. If I want to see if a lens is any good, I'll go to Phillip Reeve's review site, one of the best IMHO.
 
I really don't think Monet or Picasso or Manet or Matisse worried about this. I also don't think Henri Cartier-Bresson or Vivian Maier or Joel Meyerowitz or Garry Winogrand worried about it. I don't even think Ansel Adams thought about it this much in terms of minute degrees.

BUT .. I don't want to discourage you from it because maybe your attention to fine detail and the process of your art form is what sets you aside from all the other greats. Who knows.
 
Yeah it's not going to make or break an image. I asked this question as a hobbyist mathematician, not a photographer. Figured a photo forum might have someone who has considered these things before and could give some pointers.
 
Phillip does good reviews but he always uses a Sony camera to study M lens performances. Absolutely drives me nuts :D
Not a problem for me because I don't have a Leica. Still don't know if that should make me sad or happy. :confused-95: But I have to agree, testing a Leica lens on a Sony FF camera doesn't make much sense for people who want to know how an M-mount lens will perform on a Leica. Especially wide-angle M lenses perform significantly worse on a Sony body than on the Leica's they were designed for.
If he consistently uses the same Sony camera to test every M lens, then I suppose that's acceptable.
All lenses are tested with a Sony FF camera. In earlier days some of the testers used a 12 MP A7S, which would make me unhappy. Nowadays they mostly test with the 42 MP A7R2; and of course in the future some people will get frustrated because they probably won't redo all their tests with a 61 MP A7R4. Life sucks.
 
Back
Top