10-06-2012, 09:45 PM
[quote name='joachim' timestamp='1349467675' post='20515']
There is a proper industry standard. Focal length holds for the centre. It is perhaps most obvious for 15mm equivalent lens, which one can buy in two varieties: fisheye and rectangular. If you correct for rectangular it has 110deg, while when produced as a fisheye it gives you about 180deg. If look for details in the centre, they are same size because the focal length is the same. In short focal length is not defined via the angle of view.
One can actually view this, as if a proper rectangular lens has longer values for the focal length when you move into the corners. To illustrate this, take an old fashioned frame-viewfinder. This you can construct by holding a slide mount (the little frame you put a piece of film in when wanting to project it) in front of your eye. The distance from your eye to the plane of the mount corresponds to the focal length. Obviously this distance is larger in the corner than it is in centre.
So you can view a barrel distorting lens (like the Oly 12mm) as one whose focal length doesn't grow enough when moving into the corner.
I hope this helps.
Joachim
[/quote]
Hello Joachim,
Thank you very much for your extensive answer! It certainly helps, but I do not yet fully understand it. I had some physics long ago, but not enough to easily understand this. Please bear with me...
Your example of the angle of view with a rectangular and fish-eye shows indeed that there is a different angle of view depending on how the lens is corrected, which is more or less independent of (but limited by) the focal length. Still, with a 12mm lens that is perfectly rectangular-corrected, the field of view on a m43 would be 84deg, and all lines in view would be perfectly straight.
Is it true that when you correct an image from a lens like the Oly 12, which has a field of view of 84 degrees, that after correction, this lens, which of course still has a focal length of 12mm and an fov of 84, will generate an image that has a smaller fov? As you can see [url="http://www.opticallimits.com/olympus--four-thirds-lens-tests/414-pana_1445_3556?start=1"]here[/url] the correction of an image can take quite a bit of the fov.
Then, hypothetically speaking, this corrected image, with a fov of (lets say) 80 degrees, could be generated by a perfectly-rectangular-corrected lens of a slightly longer focal length. This would not be the same image, but would cover the same scene... Does this make any sense or am I way of the mark? Lets say you have to choose between a 17-55 lens with heavy distortion and a 18-55 lens with no distortion, and all you do is take architecture shots? My question is which lens would give a bigger range in corrected fov. Subsequently I wonder whether it is possible to measure this corrected fov and somehow include the corrected fov in reviews.
There is a proper industry standard. Focal length holds for the centre. It is perhaps most obvious for 15mm equivalent lens, which one can buy in two varieties: fisheye and rectangular. If you correct for rectangular it has 110deg, while when produced as a fisheye it gives you about 180deg. If look for details in the centre, they are same size because the focal length is the same. In short focal length is not defined via the angle of view.
One can actually view this, as if a proper rectangular lens has longer values for the focal length when you move into the corners. To illustrate this, take an old fashioned frame-viewfinder. This you can construct by holding a slide mount (the little frame you put a piece of film in when wanting to project it) in front of your eye. The distance from your eye to the plane of the mount corresponds to the focal length. Obviously this distance is larger in the corner than it is in centre.
So you can view a barrel distorting lens (like the Oly 12mm) as one whose focal length doesn't grow enough when moving into the corner.
I hope this helps.
Joachim
[/quote]
Hello Joachim,
Thank you very much for your extensive answer! It certainly helps, but I do not yet fully understand it. I had some physics long ago, but not enough to easily understand this. Please bear with me...
Your example of the angle of view with a rectangular and fish-eye shows indeed that there is a different angle of view depending on how the lens is corrected, which is more or less independent of (but limited by) the focal length. Still, with a 12mm lens that is perfectly rectangular-corrected, the field of view on a m43 would be 84deg, and all lines in view would be perfectly straight.
Is it true that when you correct an image from a lens like the Oly 12, which has a field of view of 84 degrees, that after correction, this lens, which of course still has a focal length of 12mm and an fov of 84, will generate an image that has a smaller fov? As you can see [url="http://www.opticallimits.com/olympus--four-thirds-lens-tests/414-pana_1445_3556?start=1"]here[/url] the correction of an image can take quite a bit of the fov.
Then, hypothetically speaking, this corrected image, with a fov of (lets say) 80 degrees, could be generated by a perfectly-rectangular-corrected lens of a slightly longer focal length. This would not be the same image, but would cover the same scene... Does this make any sense or am I way of the mark? Lets say you have to choose between a 17-55 lens with heavy distortion and a 18-55 lens with no distortion, and all you do is take architecture shots? My question is which lens would give a bigger range in corrected fov. Subsequently I wonder whether it is possible to measure this corrected fov and somehow include the corrected fov in reviews.