Skip to content →

Method: Correcting perspective in the darkroom

Rule number 1 of using wide angle lenses is to keep the film plane parallel to the subject. This rule, however, is more often than not promptly broken, as it is impossible to get the composition you want without tilting the lens up or down. With the lack of tilt- and shift movements on the camera, we have to make a compromise here between the composition and accepting that lines, which used to be nicely parallel, are now all of a sudden converging. Without breaking the bank, the digital photographer just has to dive into his or her favorite photo editing software and can compensate for this in a matter of minutes, albeit at the cost of a tighter crop and a slight loss of resolution. For the analog worker, this is of course more involved, but can be achieved to the same effect, if the enlarger allows both the film plane and the lens plane to rotate independently. In today’s post, I will explain how to do this, and give you an example from my own trials.

Converging lines?

Before getting technical, lets consider the example first. The photo in Figure 1 was shot near the San Jose McEnery Convention Center in San Jose, California, USA when I was there in February, 2016 for work. The construction workers nicely lined up to pass supplies from one floor of the scaffolding to the other. As you can see, the sides of the buildings converge and the building seems to be tipping backwards. This is a typical example of a situation where it was impossible to get the shot while keeping the film plane parallel to the building. At least it was from my vantage point.

Figure 1: Builders in San Jose, CA, USA. Full frame reproduction of the negative.
Figure 1: Builders in San Jose, CA, USA. Full frame reproduction of the negative.

Theory

I found the guidelines for doing this correction in Tim Rudman’s “The Photographer’s Master Printing Course” and in C.I. and R.E. Jacobson’s “Developing – The Negative Technique”. Below I summarize the procedure and add my comments.

The correction requires us to tilt the base board. Doing so allows for the lines to become parallel again. The required tilt angle greatly depends on the required correction. This is called keystoning. Getting this right takes time and probably quite some practice. For the example photo below, it took me roughly 20 minutes to get the keystoning just right. I had to make do with some books for supports and these slowly shifted over time. I ended up supporting them with a heavy box of washing powder to keep them from sliding. Some more expensive enlargers, especially those meant for professional printers, have more sophisticated and built in movements. Figure 2 shows a photo of my Durst M601 half way the procedure. I used a pile of novels to achieve the required tilt.

Figure 2: Photo of the enlarger during adjustments. A severe tilt adjustment of the base plane was needed.
Figure 2: Photo of the enlarger during adjustments. A severe tilt adjustment of the base plane was needed.

If the required tilt is small, the depth of field of the enlarging lens may be sufficient to keep the entire frame in focus. If not, reducing the aperture may solve the issue. Note though, that every lens has a sweet spot for its sharpest aperture and it is seldom at the smallest aperture available. In general, we will have to correct for the changed image plane orientation by using the Scheimpflug effect to our advantage. A detailed discussion of this effect is outside the scope of this article, but it relates the tilt of the plane of focus to the tilt of the lens plane.

Using this effect, we need to (1) tilt the film plane and (2) tilt the lens plane for it to be horizontal again. On my enlarger, both movements are possible along one axis of rotation, allowing me to correct the perspective and focus along one plane.

After correcting the focus, there may again be need for another correction. As the intensity of the light projected by the inverse square law, the part of the frame that is closest to the lens will receive a larger intensity than at the other end. If the tilts are small, and the enlargement large, i.e., the lens is far from the base board, we may not see a difference in exposure. However, if the enlargement is not so large, or if the required tilt is large (see the example), we have to correct for loss of intensity by applying a gradual increase in exposure. Getting a perfect exposure is nearly impossible though for two practical reasons. First, we need to cover the print partially and make sure that at the end of the additional exposure we precisely end at the edge of the frame. Secondly, to correct for the quadratic fall off of intensity with distance, we would have to change the speed at which the mask moves dependent on how far we are in the process. Although the first one can be achieved via trial-and-error / mastery, the second one can be considered nearly impossible to do consistently.

Results

After tedious attempts to get the tilt just right, the focus corrected and the exposure compensated, I ended with the result shown in Figure 3. Especially the final exposure was difficult to get right, although the it is not that apparent in the final print.

Figure 3: (left) effective crop; (right) final print.
Figure 3: (left) effective crop; (right) final print.

As you can see, a rather severe crop was needed to achieve the final frame. Because of the large base board tilt that is required, the crop is tapered. For those familiar with the digital workflow will recognize that exactly such a  distortion correction is also required in software.

As you can see in the example photo below in Figure 3, I would also have to correct the perspective along the second axis. The vertical lines are now nicely parallel, but the horizontal lines still converge slightly towards the left. This is unfortunately not possible in my case.

The final print still needs some work. It made a good exercise though and a useful skill to have. If I have to do this more often, I will have to make a simple tip/tilt table that can be easily fixated into a set orientation without drift. And who knows, whenever I get wealthy, I might just get a set of tilt-shift lenses to avoid this issue at the point of exposure.

Published in Methods Section