Method for Eliminating Horizontal Convergence?

[paulwainwrightphotography]

I photograph a lot of buildings here in New England that have white

clapboards. When composing a straight-on photograph of such a building, I

like to (try to) eliminate convergence in the horizontal lines of the

clapboards. Does anyone have a simple procedure for doing this that works for

them? For vertical convergence, it's simple to use a level on the back, but

it's not so easy for horizontal convergence, since it's not obvious when I've

got the camera back exactly parallel to the building. A wide angle lens makes

the problem more obvious.

 

Many thanks!

[john_stafford2]

Are you really using a view camera?

 

Correction of vertical convergence requires front rise/fall.

 

To correct the horizontal equivalence you use swing.

[ben conover]

Hello Paul, I'm just a beginner on large format, I enjoyed reading your Bio and seeing your photos.

 

Not sure I understand your dilema here. Perhaps using spirit levels on the camera, and using the grid on the focus screen to align with the horizontals in the scene.

 

Cheers.

[walter_foscari]

"...but it's not so easy for horizontal convergence, since it's not obvious when I've got the camera back exactly parallel to the building..." You said it yourself: establishing exact parallelism with the horizontal dimension of the building is not easy. I usually take a few steps away and eyeball it. You should also ensure that the camera is level (horizontally) and verify that a horizontal straight line (near the top or bottom edge of the image) properly lines up with you gg grid. If you scan your negs you can make small corrections in PS, but if you print your negs the traditional way it's a problem that is hard to correct.

[john_stafford2]

When I suggested using swing, it was because I thought your situation was one where you could not put the focal plane perpendicular with the horizontal building front.

[leonard_evens]

I've spent some considerable time thinking abou this problem. What you want is some method to determine when the leveled, plumb back is parallel to the building facade. Unfortunately, there is no simple foolproof way to do this. It is helpful to have gridlines on your gg, When you set up the camera you should check with a level that the grid lines are actually horizontal and vertical, as they should be if everything is leveled properly and the gg was installed correctly.

 

If the facade has lines which should be horizontal, you can check these against the horizontal gridlines. You should do this at least at two levels on the facade, but in a pinch, if everything else is perfectly leveled, one will do. If you can pan the camera from side to side, it is easier to see just when the grid lines are parallel to the building line.

 

Unfortunately, "horizontal" lines on some buildings actually slope one way or another.

 

If the facade is a rectangle or has a large rectangle within it, you can check on the gg the lengths of the diagonals with a mm scale. They should be equal. You can also check the heights on the gg of structures on opposite sides of the frame which should be the same height on the building facade.

 

Unfortunately, you often can't be sure the "rectangle" is actually a rectangle or elements at the same height actual are at the same height.

 

If you can identify the center line of the building, you can measure the horizontal distance on the gg from the image of that center line to the images of the building sides. They should be equal.

 

One some occasions I've actually walked off distances to find the center line of a building, but that is not always feasible. Structures such as doors or windows which should be in the center may not actually be there.

 

If there are no reference points on the building which you can identify, you are out of luck. For example, there are infinitely many trapezoidal shapes making various angles with the camera back which could produce the same image rectangle on the gg. There is no intrinsic way to differentiate them.

[jeff bishop]

I'm new at this myself, thus not an expert by any stretch of the imagination.

 

Put the camera level.

Focus on the building.

Raise the front slowly but steadily while watching to correct convergence.

 

Use a loupe. Make sure you are seeing it perfectly.

If you have lines on your ground glass, it becomes very easy.

 

If you don't have lines, check out the website for satin snow. They have print outs that will help you put your own lines on your glass.

[leonard_evens]

Sorry,

 

I made a mistake. The equal diagonals method won't work. The diagonals of some trapezoids are equal. But the equal heights on the sides method does work.

[leonard_evens]

Sorry,

 

Yet another correction. The purported horizontal line in the building facade has to be above or below the lens axis. But this is almost always the case. The further it is from the lens axis, the better.

[conrad_hoffman]

Gridlines are almost essential. Either pencil them on the inside of the gg, or print some out on transparancy material and stick that over the gg. I heard a very useful rule that went something like, if it looks wrong, it is wrong. Even if you did everything right. Just adjust the back until the geometry is correct, then adjust the lens until the focus plane is where you want it. The exact method depends on the style of camera you're using, but IMO, it's easy to make LF photography far mor difficult than it really is.

[paulwainwrightphotography]

Thanks, everybody! You've told me things I know already. I was hoping for was something like this: If the lines in the photo converge toward the left, do this... a simple step-by-step procedure that would eliminate much of the trial and error. Any thoughts?

[ben conover]

Why not invent a computer controlled grid interface that will alight the focal plane considering depth of field to any analogue camera? You'd make loads of money and lose the need to estimate visually.

 

Perhaps there is a method of estimating the amount of vetical and horizontal camera adjustment needed. Perhaps you better just guess and call it art.

 

Cheers.

[john_stafford2]

Ben, you are being facetious, no? Or do you want to know how silly that idea is. :)

[ben conover]

Sometimes silly ideas make photos, just lighten up a bit. I'd rather have a silly idea than no idea at all, and like I said, I'm just a beginner.

[ole_tjugen]

Use a compass.

 

Take a reading of the orientation of the wall, then adjust your camera back to the same direction. This is of course a lot easier if your camera contains no steel. Removing a spring back can make a lot of difference...

[michael_alpert1]

Paul,

 

Unfortunately, there are no shortcuts. You just need to work until the building looks right on your groundglass. I hope your camera has shift. Otherwise, you will need to move your whole camera for even small adjustments. If your camera does not have shift, think about a replacement. To find the center of a building, I look to the roofline for help. The slope of the roof at the sides or the way the wooden gutters angle back are often good indicators of where you are in relation to the building's center.

[leonard_evens]

"If the lines in the photo converge toward the left, do this."

 

I will assume you mean they converge to the left in the scene, which would mean they converge to the right on the gg. Just pan to the right until you eliminate the convergence. Then shift to center the image where you want it. (As has already been noted, without shift capability, you will have to move the camera and relevel everything.) Am I missing something? Are you referring to something subtle?

[john_stafford2]

Paul. I don't quite understand why this is complex. Do you mean that the building front looks like the picture below, possibly because you cannot step far enough away, and you want to diminish the horizontal conversion somewhat?

[paulwainwrightphotography]

Leonard -- Thanks, that's what I was looking for.

[joe_harrigan]

John answered this in the first post. swing.

[dhbebb]

The use of swing may help to get a receding subject all in focus, it will NOT repeat NOT eliminate horizontal convergence or correct perspective in any other way. Just as vertical shift (aka rising front)helps correct vertical convergence, side shift works in the horizontal plane. It will of course be much easier to check that you have applied enough correction if you have a gridded ground-glass screen.

[william_moss]

There is a simple technique that hasn't been mentioned here. Swing is definitely NOT the answer- as mentioned, it just allows both near and far portions of the building to be in focus.

 

The diagram above is excellent. Ideally one would want to move to a position where one can face the middle of the building directly.

 

If that's not possible, the answer is SHIFT, not SWING. If you shift (either front or back standards, or both) enough to get all of the building in the image, while still keeping the standards parallel to the plane of the building's facade, you're home free. This is the same technique used, for example, to shoot into a mirror without getting a reflection of oneself. See any of the Stroebel books for more info on this. Or check out the (brief but informative) summary on "View Camera" on Wikipedia.