A zoom lens is not a panacea, but a way out.

A zoom lens is not a panacea, but a way out.

I will immediately stipulate that in this case, the term «zoom lens» means a varifocal lens (with variable focal length) with servo control, since the term «manual zoom lens» exists and has the right to exist.

This magic word is ZOOM. And it is written entirely in capital letters in the price lists of some companies. It is followed by the number of multiplicity — 6 (so-so), 12 (already better), 20 (great!), 34 (cool!), etc. True, the price also changes several times. But if the customer has been persuaded to spend money, then… And here, instead of the apparent solution to all the problems, new ones suddenly begin to appear, one after another. And the cooler the zoom lens (the longer its maximum focal length), the cooler the problems. Therefore, we will talk specifically about long-focus zoom lenses.
Before considering the specifics of a long-focus lens in general and a long-focus zoom lens (I will allow myself to call it that, meaning by this term a zoom lens with a maximum focal length of somewhere around 150 mm or more), and giving recommendations on use, I suggest that the reader (consumer, installer, etc.) try to answer a number of questions on their own.
Why is the most common lens standard on the market 3.5–4.5 mm; these are the lenses that almost all frameless cameras come with “by default”?
Why does Hollywood build practically railroad tracks and special boom cranes on film sets to move cameras, which cost hundreds of times more than the most zoom lenses? And how tiresome is it to watch amateur videos when the cameraman is engrossed in continuous work with the zoom lens and does not stop filming when the zoom changes? Why, for example, does the BBC company in «Wildlife» film all sorts of snakes, crocodiles, hippos practically point-blank, and not from a distance of 200-300 m? Was there no money for a decent zoom lens?
Why aren't there «mass market» optical sights with a magnification greater than 8? (It would seem that if we made a 40-50x sight, everyone would become a sniper, without having to do very expensive special training for people.) There are many more questions like these that can be thought up, and they are all directly related to long focal lengths, variable focal lengths, and the problems associated with them.
Any decent manual on film photography, reference books on lenses notes (at least, it used to be noted) that the resolving power of a zoom lens is always worse than that of a fixed lens. And this is understandable. It is quite difficult to make a high resolving power of an optical system that includes lenses that are not fixed in one strictly defined position. So, if the resolution of a decent photo lens with a fixed focal length is about 50 lines/mm in the center and 40-45 at the periphery, then a zoom lens with a resolution of ~ 28-35 lines/mm in the center is a good zoom lens. This will have a relatively lesser effect on the resolution of an analog video camera, although this should also be kept in mind, especially with long focal lengths of zoom lenses, when the difference in resolving power from the center of the lens to its periphery begins to rapidly fall (the object of identification must be «driven» into the very center of the screen).If we talk about megapixel cameras that are rapidly filling the market, then neglecting the lens resolution can negate all those advantages for which not only the megapixel camera itself was purchased, but also the entire IP system was built. It is no coincidence that a separate term “megapixel lens” has even appeared on the market. By the way, the term is quite loose. The lens resolution has a very specific numerical value, and not just two gradations — “mega” and “not mega”. Incidentally, I will note that I personally have not seen such a characteristic as the lens resolution in any advertising (or non-advertising) information on lenses on our market. It’s a pity. It would be possible to compare not only by price.

Fig. 1. Dependence of depth of field on focal length.

The second problem of long focal lengths of a zoom lens – a small depth of field – concerns us directly. Of course, we can simply declare that the longer the focal length, the smaller the depth of field. However, what is simply declared is often simply forgotten. And since this point is quite important, it makes sense to illustrate it clearly (Fig. 1).
There is a certain object AB. There is a lens with variable focal length. At focal length f1 we have an image of the object A1B1, at focal length f2 – A2B2. f2 is clearly (approximately 3 times) larger than f1; naturally, the scale of the image A2B2 is significantly larger than the scale of A1B1.
In video surveillance systems, observation is mainly carried out for moving objects. Let us imagine that our object AB has shifted in the direction of the optical axis of the lens (has come closer) by a certain distance d (position A−B−). Let us construct new images of object A−B− at f1 and f2 – A1−B1− and A2−B2−, respectively. Image A1−B1− has shifted from its initial position A1B1 by a distance d1, and A2−B2− has shifted relative to A2B2 by a distance d2. To obtain a clear image of the object, the matrix of the video camera must also be moved (focusing the image) in the first case by a distance of d1, and in the second — by d2. Obviously, in this case d2 is significantly greater than d1. It is possible that in a particular case focusing for f1 will not be required at all (for example, the lighting conditions will cause the aperture to close, which will completely provide the necessary depth of field). For the focal length f2 in this case this is hardly possible. Focusing will have to be done.
If the object is located, as they say, at infinity (for long-focus lenses this distance is ~ not less than 200 f), the image is located in the focal plane, and when the object moves at distances greater than «infinity», no additional focusing is required. In all other cases, almost continuous sharpness adjustment is required as the object moves, and the closer the object is to the lens, the greater the need (let me remind you of the basic lens formula: l/d1 + l/d2 = 1/f, where d1 is the distance from the object to the optical center; d2 is the distance from the optical center to the image plane (matrix); f is the focal length of the lens). So if you are trying to examine the face of a person entering a yard gate at a distance of 40-50 m from the camera with a long-focus zoom lens (for example, with a focus of 200 mm), or read the license plate of an entering car, it is quite difficult to do this.
In fairness, it should be noted that the majority of zoom lenses with long focal lengths are installed at distances to the object that fit into the concept of «infinity», the problem of depth of field is not so widespread. But it makes sense to always remember this and understand the physics of the process. The next point that should not be ignored is the lens aperture, which is determined by the relative aperture — the ratio of the effective diameter to the focal length 1/F. In practice, they operate with the inverse value of F.
The larger this figure, the less light flux will hit the matrix.
As for the zoom lens, the sensitivity of the video camera is specified for a specific F value. Naturally, the zoom lens has maximum aperture at the minimum focal length. If this value does not match the relative aperture value for which the sensitivity of the video camera is specified, it can be easily translated for a specific F value. But with an increase in focal length, the F value will increase proportionally to the increase in focal length, and the sensitivity of the «assembled» video camera will accordingly decrease. The zoom lens ratio is the same as the sensitivity decrease at maximum focal lengths.
Now about the most critical point. In 99 cases out of 100, long-focus zoom lenses are designed to work at distances often amounting to hundreds of meters. In 99.9 cases out of 100, they are installed on rotary devices. Control, naturally, is from the remote control. Feedback is through the image on the monitor. And in 99 cases out of 100, the customer assigns special, and sometimes unique, tasks to the zoom lens, far from the tasks of simple object detection. Quite often, these are tasks, for example, identification of a person. Let's try to analyze this situation.
Let's assume that from a distance of 100 m it is necessary to identify an unknown person to the extent necessary for his/her subsequent unambiguous identification (a real-life task). That is, it is necessary to provide a field of view of ~2 m. We get that for a matrix format of 1/3 inch, the focal length of the lens is ~200 mm. The viewing angle of such a lens will be 1.35°. It makes sense to think about these figures.
A rotation of the optical axis by 1′ (one arc minute) is equivalent to a movement of the observed object by 2.5 cm. If the optical axis oscillates with such an amplitude, then each point of the object is transformed into a spot with a diameter of 2.5 cm. There can be no sharpness of the image in principle.
And any immobility is always relative. Are you familiar with the situation when dishes rattle in a cupboard at home while piles are being driven in somewhere nearby? Or is there a railway nearby? Or is there a subway line running under the house? It is quite enough for the shaking of our long-focus lens to affect the sharpness of the image. If the camera is installed on a mast, the latter always has its own vibrations, the amplitude of which is greater the higher from the base. Additional fastening is always required.
In the vast majority of cases, such cameras are installed as all-weather ones, i.e. in a fairly large hermetic housing, on a rotary device. With the windage of the hermetic housing, even a small wind is enough to cause its vibrations within the play of the rotary device. Now take any turn signal of any company and feel the play with your hands in the planes of its rotation. In most cases, it will not be minutes, but degrees. For the case under consideration of identifying an unfamiliar person from a distance of 100 m, 1.35 ° is a complete departure of the frame from the screen. And even the steering wheel of a Mercedes has play. The level of execution of the mechanics of those turn signals that are available on our market obliges them to have play. You can choose it, but at the same time you can expect mechanical jamming at any time.
Modern consumer and professional video cameras equipped with long-focus zoom lenses have a shake compensation circuit in their electronics, called a hand-shake compensation system. And although such a circuit is nothing particularly unique, I was unable to find cameras with shake compensation «head-on» on our market. Too bad. This issue is quite relevant. By the way, our organs of vision have such a circuit in their composition, and of very high quality — we can clearly see the situation in the rear-view mirror shaking while driving. But on the monitor from a shaking camera — alas. The situation will look even worse in the recording.
The circuit for direct control of the rotary device (switching control voltages directly to the motors of the device) will entail additional serious difficulties for the operator.
In the theory of shooting at moving targets, there is such a thing as a “personal error” – the time from the brain issuing a command to shoot until the trigger is actually pulled. So, depending on the reaction of a specific individual, this time is from 0.15 to 0.35 s. The majority of rotary devices have a horizontal rotation speed of 6 deg/s. If we assume that a person with a minimal personal error is sitting at the operator’s console, then in 0.15 s the optical axis will rotate by 0.9°. For the case considered above (f=200 mm, distance – 100 m, horizontal field of view – 2 m), the miss will be 1.33 m (67% of the field of view). That is, the operator has already turned off the rotation (in fact, it seems so to him), and the rotation continues. Let’s add some more backlash, and it may well happen that the required scene completely disappears from the monitor screen. The ability to “hit” at long focal lengths is the result of practical training. It is necessary to aim the zoom lens, so to speak, by the method of successive iterations (approximations), starting from a wide angle. For this purpose, a kind of sight is often applied to the monitor screen (some draw a dot, some — a cross), corresponding to the optical center of the lens. It is good if there is an opportunity to change (reduce) the rotation speed when working at a «long focus». When organizing control via telemetry receivers, this can be done by switching to a step-by-step algorithm. The main thing is that the customer understands all this, and does not rush to accuse the installer of foisting off low-quality equipment.
As a lyrical digression. The customer wanted to observe a very large area (horizontal angle of 60 degrees) from a distance of 100 meters with a single camera. The small scale is obvious. In response, I was asked to «first enlarge the image, and then expand it» so that the entire necessary field would be on the screen in the required enlarged scale. So it is better to tell the customer in advance everything he does not know, than to sort things out with him for a long time later.

Fig. 2. The effect of changing focal length on depth of field

By controlling the «scale» function, pressing the buttons on the remote control, we change the focal length of the lens. It is good to imagine how this affects the sharpness of the image. Let's consider the constructions in (Fig. 2) for the general case when the object is not at «infinity». d (with d positive), and only if the object is at «infinity», will we have the equality of these values. In practice, this means that with an increase in scale, a transition to long focal lengths, the sharpness will go away more and more rapidly. Often, such a transition has to be done in several stages, focusing the image at each of them. Otherwise, even the general contours may disappear, and it becomes unclear where the camera is pointing and what we will see after focusing. There are and are becoming more and more widespread zoom lenses with automatic focusing. But you should always remember that any «automatic» has a certain «play» of values, a certain error. The smaller it is, the higher the price of such equipment. If you decide to save money, you will have to take into account this «automatic» possible focusing error.

So, conclusions.
Always, if circumstances allow, a lens with a fixed focal length is preferable to a zoom lens.
Always, if circumstances permit, the closer the camera is to the object of observation, the better: the shorter the focal length with the same field of view, the larger the viewing angle, as a rule, the higher the lens resolution, the smaller the focusing error, the greater the depth of field, and ultimately the greater the recognizability of the object. Perhaps it makes sense to install a separate support for a stationary camera, bringing it closer to the object of observation, rather than playing up the distance with a long focus. Incidentally, regardless of video surveillance: when you take pictures with some wonderful «point-and-shoot», use your feet more often to place the scene in the frame, and less often — the zoom lens.
The identification problem does not necessarily have to be solved every time within the framework of the general video surveillance system. Perhaps a separate solution will be more effective.
If you still cannot do without a zoom lens, then you should perceive it not as a panacea, not as a universal solution, but as a way out of the situation and always a compromise. The main thing is not to promise the customer too much, so that you don’t have to fight the objective laws of physics yourself. They will win anyway.