I think the best way to find out the answer to that question is take a few pictures with the different lenses at appropriate distances and find out for yourself.
I will surely do that...
I didn't mean for that to be a sarcastic answer. Really, the feature is very new, and a lot of others like yourself are trying to understand what it will really do. And while you can ask others if it does certain things, and get their opinions about it, the only way you can really know how it will work in your situation is to try it.
I understand completely. My question was really on the mathematics of the
program. What I wanted to know was whether the objective was to straighten
out the optical distortion in a wide angle lens so that it functioned as a
wider version of a normal lens. It seems in playing with the program that this
was the case.
Yes, the Canon 24-70 profile will minimize residual barrel distortion at 24 mm at this distance. (The profile should actually work well at all focus distances.)
Keep in mind, however, that only optical (i.e., lens) aberrations are corrected by the profile. Other types of distortion, such as stretching of objects that happens towards the edges & corners of images when taken with wide angle lenses, are out of the scope of the profile.
Effectively, the purpose of the profile with respect to distortion is to take straight lines in your scene and render them as straight lines in your photographed result. Whether or not this is desirable will depend on your image content. This is one of the reasons why there is a "Correction Amount" set of sliders. For example, you may find that you wish to correct some, but not all, of the lens distortion. In this case, you could (for instance) set the Distortion slider to a value of 50 (instead of the default of 100) to do a "half-way" correction. So you would get something in between no correction and full correction.
Does this help?
Hi Eric...that does help...thanks...I will lay down a square or rectangle
on the floor and shoot from a ceiling mount so that I can actually measure
the amount of correction and distortion. If the lines become straight why is
there still distortion present?
Would I be better off then using a 35mm fixed lens rather than a 24-70
Eric...I am not able to open CS5. It opened once after I updated Camera Raw
to 6.1 but it will not open now.
I am running a Mac 2.66 GHZ Quad Core with 6 GB RAM and version OSX 10.6.3
There are many types of distortion that can be present in an image.
The type of distortion that is corrected by our new lens profiles is an aberration of Seidel sometimes called "curvilinear distortion." When using a theoretically perfectly rectilinear lens, a straight line in the scene would get recorded as a straight line in the image. In practice, this does not always happen, and we commonly encounter barrel distortion in a zoom lens at the wider focal lengths, and pincushion distortion at intermediate & longer focal lengths. With some lenses, more complex curvilinear distortion is present.
However, there are other types of distortion that can be present, which are not the fault of the lens. For example, if you shoot a group portrait with a 14 mm rectilinear lens, then the persons may appear unnaturally stretched (esp at the edges & corners of the image), even if the lens was perfectly rectilinear. The apparent distortion that yields unpleasant stretched heads & bodies is simply due to perspective projection and is unavoidable in rectilinear lenses. In such cases it may actually be desirable to have some barrel distortion present (and you can accomplish this to taste in Camera Raw 6.1).
It is somewhat analogous to the mapmaker's problem. You cannot create a map of the earth on a flat sheet of paper without distorting something. You can optimize for one set of characteristics, but you give up something else.
I can give you some simple example images to see, if you are interested.
In the meantime, you can see an example here (bottom of page):
Zeiss also has a very informative article on distortion (including an example of the stretching at the very end):
Thanks. Very helpful I will shoot a large rectangle tomorrow with the
35,50 and 24-70. If you want I will send the raw files. It should be
clear how the program is functioning.
Sent from my iPhone
Awesome explanation, Eric. Thanks!
Hi Eric...I did my experiment...the situation is this...the camera will be
about 8 feet from the subjects mounted facing directly down. The subjects
will be five dancers laying on the floor intertwined. I will need probably at
least six feet by six feet for the shot.
Testing the scene at this distance it seems to me that I may be able to get
by with the fixed 35mm 1.4 lens...but may have to use my 24-70 zoom. I
tested the barrel distortion and the lens correction programming. It seems that
I can make either of these lenses work with the amount of control the
program is giving me.
Theoretically I should get better results with the fixed lens...any
I also was wondering about this topic for my NIkon lenses; thanks Eric, for the link to the Zeiss article on distortion, this is what I needed in the other thread when I had asked for links to discussions on just what is "correctable" distortions vs non-correctable issues, etc.
Good luck, Roger, with your shoot with the wides,
To add to Eric's great explanation, the corrections here are for barrel and pincushion distortion. That can be present in a lens of any focal length.
What isn't corrected is the effect in wide angle lenses where there is spreading of the image of three dimensional objects away from the center of the field. It happens at any distance, not just close up. It is caused by the extreme angle of projection from the back of the lens to the flat film/sensor plane. Think of a straight on photo of a building that has equal size columns across the facade with a wide angle lens. The columns at either end of the image appear wider than the (same width) columns in the middle. Retrofocus design wide angle lenses help a little bit (the rear of the lens is further from the focal plane than in a standard wide angle), but the distortion is usually there.
There is also distortion from being too close to the subject. Focal length doesn't really matter in that, except that longer lenses don't have the field of view to take in the close subject and can't usually focus that close. So we associate this distortion with wide angle lenses. This sort of distortion would be that where a person's nose in the middle of the frame looks very large and there is quick falloff in size to the eyes, ears, background. I think of this as more an issue of perspective, and this is controlled by camera placement, not by focal length or lens design.
Another good page on Paul Van Walree's site is http://toothwalker.org/optics/misconceptions.html