Perspective Distortion

[terry_long]

Correct me if I'm wrong...please.

 

I shoot mostly landscape/scenic photography. However, fall is

getting very close and if I want to isolate a certain tree, I'll

have to get pretty close (for 4x5, the largest focal length lens I

have is a 210mm) which means I'll have to tilt the camera upward to

encompass the entire tree. If I tilt the rear standard so it's

roughly parallel, vertically, to the tree and use the front standard

for focusing, can I surmise I won't have a large amount of

perspective distortion?

 

Thanks in advance.

 

Terry

[michael_briggs2]

When the front and rear standards are parallel, perspective is rendered just as you see it -- it you looked in the same direction and somehow traced the scence on tracing paper, it would have the same perspective as the camera. However, it is a convention, at least of Western Art, that if the sight line doesn't greatly deviate from horizontal, then the perspective should be rendered with the tracing paper or film plumb, i.e., perpendicular to the ground.

 

So when you keep your rear standard plumb, then you will obtain the perspective rendering that is generally regarded as pleasing / standard / artistically correct. Assuming that your tree is vertical, this is what you will be doing.

 

If you tilt the camera upward and bring the rear standard to plumb, but not the front one, you plane of best focus will tilted with respect to the film. If you want the plane of best focus to also be plumb, then you need to bring both standards to plumb. Geometrically, this is equivalent to using either front rise or rear fall, or both. If your camera has sufficient front rise / rear fall, these movements will probably be easier to use.

[leonard_evens]

I don't have anything of any substance to add to what Michael said.

 

But perhaps here is a way to think about it which may clarify things a bit. The only thing that matters to the optics is the relative positions of the two standards. The supports of the standards and the bellows are there to hold things in place and block out light, but they play no role in the optics. You can accomplish what you want two ways---or by some combination---subject to the physical limitations of your setup. You can use a rise/fall of the standards. Or you can tilt the camera up and then bring the standards back to plumb and parallel. It doesn't make any difference how you do it; it is only the final relation that counts.

[ellis_vener_photography]

while Michael is pretty much right in the academic sense, sometimes having the back not

quite vertical, (i.e. tilted back very slightly can create a sense of height. My advice is to

take some Polaroid along with you and test both approaches.

[david l.]

Ditto Ellis' comment on the Polaroid. Also, keep in mind that converging verticals are less of an issue when photographing a single tree because there are no true verticals to begin with (the tree's trunk is tapered).

[conrad_hoffman]

With a 210mm on 4x5, and assuming you're standing on the ground, you'll be fairly far away from the average size tree to get most of it in. 210mm ain't exactly wide. IMO, the whole converging vertical issue won't come into play here. Remember too that the average tree had converging verticals long before you got there- smaller at the top than the bottom. If you're doing an angled close-up, say of a trunk area, then you might want to adjust so the edges are parallel.

[ken_lee11]

As a general practice, it's a good idea to level out the camera first: front and rear standards alike. Pointing the camera at the tree, you can then raise the front standard to center the tree as required. In this case, I lowered the lens, a 240mm Fujinion A.

 

<p><center><img src = http://www.kenleegallery.com/ludlow.jpg></center>

<p>I look forward to the cool clear air of autumn, and to the mists which form just after sunset.

[www.graemehird.com]

"When the front and rear standards are parallel, perspective is rendered just as you see it ...."

 

I must respectfully disagree a little with Michael on this point: by keeping the standards parallel, the perspective will be removed and the scene will be geometrically correct, rather than being rendered "as you see it".

 

Verticals really do appear to converge from a person's point of view, but we often use the view camera's movements to counteract this perspective.

 

As someone else mentioned, a 210mm lens is not that wide, so you won't be getting that close that you'll need to worry about perspective control very much (assuming you want the whole tree in the frame). In any case, your assumption about tilting the rear standard is correct, but to maintain focus, you'll also need to tilt the front standard by a similar amount so that it too is (nearly) parallel to the tree. If you're too close, you might exceed the coverage of the lens. Good luck, and show us the pic when you're done!

 

Cheers,

[uli_mayer]

Greame is right. One has only to look at "Old Water Tower" onw p.158 of AA's "Examples" to see how awkward things can look if vertical correction is overdone, especially when using a wide angle lens. Uli

[leonard_evens]

Graeme and Uli,

 

I beg to differ with the two of you.

 

First, let me address the issue of wide angle lenses. You certainly do get "distortions" when you use a wide angle lens even though the perspective is geometrically correct. But these result from a failure to view the final image from the same relative point of view. If you put your eye in the proper position, which may be difficult for a picture taken with an extreme wide angle lens, what you see from that position will be just what you would see in the scene.

 

Given that one doesn't usually view images from the proper point of view, particularly for wide angle images, one could argue that my point is irrelevant. But I would maintain that typically you would make things worse by pointing the camera up at an angle and having verticals converge in such a picture.

 

More generally, "what your see" is a complex question. It is not the scene projected on your retina. You scan the scene with your eyes and you move your head and your eyes. Your visual system puts together an image from that which includes many departures from geometric straight line perspective. For example, objects in the distance generally look larger to you than the geomtric image would suggest. That is called size constancy. Perceptual psychologists spend a lot of time investigating these matters, and I don't think they claim to understand them fully even now. One element in the art of photography is to work around this difference to produce plausible and pleasing images, or sometimes, to dramatise the departures from "what you see" to expand the viewer's vision.

 

As to converging verticals, I have to say that I don't see them converge when I look at a scene unless I have to turn my head up at a sizable angle. If I'm far enough away from the subject so that is not necessary, I don't see the verticals as converging. Others may differ about that. Indeed, some commentators have said that before the age of photography, people didn't usually see verticals as converging in natural scenes, but now they do because they have been trained by all the photographs they've seen with converging verticals.

[uli_mayer]

Leonard,

 

I fully agree with your statement. My point is only: We may be so accustomed to converging verticals in our visual perception that leaving a bit of it in our photos is often better than struggling for full geometrical correctness at every time. Even straight technical and architectural photos need a bit of "salt" to look right. Indeed we have a wish for "size constancy", but there may also be a wish for "perspective constancy" we want to have confirmed.

[www.graemehird.com]

Leonard,

 

Whether or not you notice vertical convergence, it's there.