Convergence of vertical lines

[Mike Lopez]

I'm still new at large format photography, and I've recently been dabbling in architectural photography (exteriors). I know that tilting the camera vertically will make buildings appear to "fall backward" if the tilt isn't accounted for by adjusting the lens and film planes. But I'm wondering what causes vertical lines to converge inward toward the frame center. And how do you eliminate the convergence? I've seen some slight convergence of vertical lines in my negatives, and it's small enough to make me think I'm *almost* doing everything correctly, but not quite. Thanks.

[David Karp]

Mike,

I can't really give you a good explanation of the optics. I can tell you what to do to avoid the "keystone" effect when photographing architecture.

Your camera needs to have front and or rear rise and fall, or you use a combination of these two movements, or you can use a combination of front and rear forward tilt to accomplish the same thing.

Front or rear rise (or fall): This is the simplest way. If your camera has these movements, level your camera and raise the front or lower the rear until you see what you want on the groundglass.

Combination of front and rear rise: If your camera does not have enough movement (and your lens has enough coverage) use both rear fall and front rise.

Forward tilt: You can also point your camera up, and then bring your front and rear standards to vertical using forward tilt. This is handy if you don't have any or enough rise (or fall).

Hope this is what you are looking for.

[Tim Curry]

Short answer. The film plane must be parallel to the surface to avoid convergence. In a tall building, the back must be kept vertical. On a fence, the film plane (back) would be rotated to match the plane of the fence.

[John Cook]

Tim is absolutely correct.

However (there is always a "however"), sometimes when you are very close to a building and a major correction is required, the building may appear to the eye to be over-corrected, even though it isn't.

We used to run into this phenomenon when photographing Merrium-Webster dictionaries in Springfield, MA. Looking slightly down on a standing book (to show the top pages), if we made the camera back perfectly plumb the book would look as if its base was wider than its top. The left-right measurements would be exactly identical, but it would still appear over-corrected. So we would slightly under-correct to make it look normal.

And thus, the art director's admonition to us, "It doesn't have to BE square, as long as it LOOKS square".

The same phenomenon happens with a telephoto shot of an automobile, in which the rear wheel always looks larger than the front wheel. Go figure!

[John Hennessy]

<"It doesn't have to BE square, as long as it LOOKS square".>

It is for this reason that the ancient Greek architects tapered columns smaller toward the top. Otherwise, a perfectly cylindrical column looks too fat at the top. It is hard to judge in the field so sometimes I bracket not for exposure but for convergence.

[John Downie]

From an article on Greek temples:

The Greeks developed an augmentation technique known as entasis to avoid an optical illusion caused by the shaft's fluting (parallel vertical lines). In a tall structure, like the Parthenon, such lines appear concave. To compensate, the Greek architects made the columns slightly convex.

http://ancienthistory.about.com/library/weekly/aa061300a.htm

[Alan Davenport]

The lines converge on the film because, when the camera back is tilted (up), the light rays from the bottom of the building must travel farther from the lens to the film (relative to the rays from the top of the building.)







Yeah, I know, it ain't very artistic... (



As Tim and John said, to avoid convergenge you need to keep the camera back vertical. In Optics in Photography, Kingslake points out that the camera is indeed faithfully recording what it sees, converging lines and all. When it is presented to a viewer in a 2D print, it doesn't seem natural because the brain doesn't make the automatic correction that happens when you look up at an actual building. If the camera is tilted up enough (30 degrees or more) that the print again seems OK. Apparently, convergence is expected when looking up that steeply.

[Ernest Purdum]

Mike, I'm not entirely sure I understand your question, When I read "vertical lines to converge inward toward the frame center", it sounds like you may be talking about pincushion distortiom. Do you mean straight lines that are not quite parallel? If so, you have received some good answers. If though, you mean lines that are not quite straight but are bowed in a little in the middle; this is pincushion distortion and is a characteristic of the particular lens. There is no way that movements can correct it.

[Mike Lopez]

Good point, Ernest. The lines are straight. I appreciate the answers I've received.

[Walter Glover]

I agree that Mike's conundrum seemed more likely to be pincushion distortion rather than good ol' covergence.

However, on the point about convergence and the difference between BEING square and LOOKING square, Sinar have always recommended that beyond a 20º inclination or declination that the convergence be accepted. To this end they incorporate a 20º detent in both the F-cameras and P-cameras tilt mechanisms.