Lens Separation in Stereo Photography
By David Lee
The purpose of this article is to help the stereo photographer choose a camera separation in practical situations.
Obtaining a Consistent Amount of Deviation
If one chooses to obtain a consistent amount of deviation on the film from one image to the next there are mathematical principles that will help. The formula seems like a complicated matter, but by the end of this paper I will explain how to simplify it so that one can easily use it in practical situations. There are five parts to the formula:
The Depth Factor is based on the amount of deviation (a measure of the difference of the distance between the far points and the near points of a stereo pair) that looks good in a stereo pair. You can see how much deviation you like in a typical stereo pair by:
The most commonly used Depth Factor among stereo photographers is 1/30, hence the 1/30th rule. The 1/30th rule in practice means you measure the distance to the near point of the scene and then divide by 30. I am aware of people using from 1/20 th to 1/60th the near point distance as a depth factor.
Near Point Factor
You simply measure the distance to the near point. Sometimes you can do this by using the distance scale on your camera. When you deal with distances over about 20 yards then you must estimate the distance in some other way. You can get either optical or laser range finders and topographic maps are sometimes helpful for really long distances
Far Point Factor
It is usually more difficult to estimate the distance to the far point since it is often very far away. Fortunately this is not really a problem because all you really need is a broad estimation of how far it is compared to the near point.
When the camera lens and the viewing lens are (approximately) the same focal length, then the lens factor has a value of "1", and so does not change the separation. When the camera lenses are shorter than the viewing lens then the deviation (amount of depth) is reduced for the same separation. When the camera lenses are longer than the viewing lens then the deviation is magnified for the same separation. Therefore, to get the target deviation it is necessary to multiply the camera separation by a factor proportional to the change in focal length of the camera lens. In other words, if the camera lens is half the length of the viewing lens then you must multiply the calculated separation by 2. If the camera lens is twice the length of the viewing lens then you must multiply the calculated separation by ½.
The Complete Formula
Here is the complete formula:
Further Analysis of the Far Point Factor
I am sure that most people would look at the Far Point Factor like it was a foreign language and give up on it before they started. In fact, I wouldn’t even use it myself while out in the field. Fortunately, when we actually examine the formula it has a pattern that simplifies things greatly:
a) If the far point is 2 times the near point, the Far Point Factor is 2
b) If the far point is 3 times the near point, the Far Point Factor is 1.5
c) If the far point is 4 times the near point, the Far Point Factor is 1.3
d) If the far point is 5 times the near point, the Far Point Factor is 1.25
e) If the far point is 6 times the near point, the Far Point Factor is 1.2
f) We start to reach diminishing returns and we can assume the factor is 1.
Too Much or Too Little Camera Separation
If there is too little deviation then the image will have little depth and will look flat. How flat is too flat is a matter of opinion. An experienced stereo photographer may have a pretty good idea whether an image might look better with more depth, but for the most part it is probably better to have too little rather than too much depth. If there is way too much depth then the image may be un-fusable, but before it becomes un-fusable it may just cause eyestrain, which may be more noticeable to some than to others. When I bracket camera separations I usually find that the most pleasing separation is not the widest that I can fuse. I won’t even necessarily feel that it is a strain, but that it just does not have as smooth a look as one with less depth. A deviation of 1/30 the focal length of the viewing lens usually looks good to me, although I can easily fuse 3 times that much deviation. I have found that many novices using twin cameras have the habit of getting more deviation than they would otherwise chose if they were to view an alternate pair with less deviation. I think that bracketing camera separation is a good idea for people who are just starting out so they can compare how images look when they have more or less depth. I would suggest using your best guess as your initial separation, and then using twice as much and half as much, just like bracketing exposure in one stop intervals.
Accuracy of Calculation
When we talk about accuracy of exposure we may say we need to be within one half stop to result in an optimum slide (for instance). We can look at depth in the same way. If one had a series of pairs the depth of which varied by 5% from one to the next, there would not be one pair that was so superior to the rest that the viewer would say, "This one is the only one that works and the others are wrong." Instead the viewer would be more likely to say, "These 3 work the best and the ones with less depth look pretty good, but they become gradually too flat, while the ones on the other side look pretty good but they become gradually too deep and are less smooth than the others." If asked to pick out the best depth, it is likely that several viewers would choose different amounts of depth or that an individual might say that he/she saw advantages and disadvantages for several of them but that no one stood out as clearly superior. In my experience making lots of images with varying depth, I have concluded that a range of one stop of depth (the greater depth having twice as much depth as the narrower one) is pretty acceptable. Another way of looking at it is to say that if there were an optimal separation, then a factor of one half stop of depth in either direction would look good. If this hypothetical separation were 1/30th the distance to the near point, then the acceptable range would be from 1/40th to 1/20th the distance to the near point. So when we are aiming for a good lens separation we do not have to be perfect, but if we get within an acceptable range then the depth will look good. We must remember that no single lens separation will seem preferable to every viewer. Don’t get either too obsessive or too sloppy about determining camera separation.
I will try to give a few practical tips to those who are still trying to figure out what to do:
Measure or estimate the near point distance.
Times 1/30 (4" per 10’ or 1’ per 10 yards or 50 yards per mile)
Times far point factor (pick one):
- 1 - if far point = ¥
- 2 - if far point is < or = twice the near point
- 1.5 - if far point is > twice the near point, but < ¥
Times _____ (put your lens factor here)
Separation = Depth Fac. x Near Point Fac. x Far Point Fac. x Lens Fac.
Separation = ________ x ____________ x ___________ x _______
Example: near point = 10’, far point = 30’, normal lens
Separation = 1/30 x 10’ x 1.5 x 1 = 6"
An Even Easier Way to Remember Things
Knowledge of very few principles may result in excellent depth in the vast majority of cases.
- Use a stereo camera for people and other close subjects.
- Use a separation of 1/30th (4 inches for every 10 feet, 1 foot for every 10 yards) the distance to the near point of the scene when the far point is at ¥ .
- If the far point is distinctly closer than ¥ then add 50% more separation.
- When using wide lenses (say twice as wide as normal), use twice as much separation.
- When using long lenses (say twice as long as normal), use half as much separation.