A digital camera faces the challenge that it has to analyze the sharpness of an object even before the shutter is released. In case that the object is not in focus yet, the camera has to find the right lens position so that the object is finally rendered sharply on the image sensor. These tasks are fulfilled by a dedicated autofocus system that consists of sensors and actuators.
Regardless of their individual designs and implementations, all autofocus systems share the same fundamentals. An autofocus sensor is placed inside the camera body where it records light that has passed through the lens. The signal generated by the autofocus sensor is transmitted towards the central processing unit (CPU) of the camera where the overall sharpness is analyzed. Any adjustment of the lens position required to focus on a certain object is performed by an actuator that is controlled by the CPU.
The distance between the lens and the focused image depends on the distance between the lens and the object to be photographed. When a lens is placed at a given distance from the camera sensor then this distance and the individual focal length of the lens determine the distance to an object that is in sharp focus. Objects at any other distance are not clearly focused.
In order to bring an object at some other distance into clear focus, it is required to either change the position of the lens or to change its focal length. Each of these changes is possible, and both are supported by modern DSLR cameras. The following table gives an idea of how much the lens position must be changed to bring different objects into focus.
Distance of the object in front of the lens in units of the focal length
Position of the image behind the lens in units of the focal length
For instance, it is assumed that a lens is used with a focal length of 60 mm. The first entry in the table above says that an object 600 mm in front of the lens (10 x 60 mm) is imaged 66.6 mm behind the lens (1.11 x 60).
If this object is to be projected on the camera sensor, the lens must be located 66.6 mm in front of the sensor. The next entry says that an object that is 6000 mm (6 m) in front of the lens will be imaged 60.6 mm behind the lens. The lens must therefore be moved closer to the camera sensor in order to focus on this more distant object. The same thing is true for the other entries – the lens must be moved closer to the camera sensor as the distance to the object increases. Note that the changes become smaller and smaller as the distance to the object increases. In order to bring objects from 600 mm to infinity into clear focus, this particular lens must be able to move about 6.6 mm – from 66.6 mm in front of the sensor to essentially 60 mm in front of the camera sensor. The figure below summarizes these lens positions.
The lens motion required to bring these objects into focus is surprisingly small, and it is not difficult to design a camera that allows the lenses to move by these amounts relative to the camera sensor. On the other hand, the position of the image changes more and more rapidly as the objects gets closer to the lens, and most lenses cannot move far enough to focus on close-in objects for that reason. For instance, an object that is 5 focal lengths in front of the lens has an image that is 1.25 focal lengths behind the lens, while an object that is 2 focal lengths in front of the lens has an image that is 2 focal lengths behind it. These motions cannot be accommodated by standard lens constructions.
In photography, most camera lenses consist of multiple glass elements to reduce aberrations that typically occur on single lenses. Nevertheless, such an array of individual lenses combines to a group with an own focal length and performs like a single lens in principle.
While it is possible to focus on an object by changing the position of the entire lens assembly, it is highly inefficient. Some professional telephoto lenses can easily weigh over several kilograms which is too much to be driven by a small actuator. It is more common to have a smaller, dedicated lens group that is the only part to be moved when focusing is performed. This procedure is referred to as internal focusing and provides numerous advantages. As the front lens does not move or rotate during focusing, any filter applied to the lens does not rotate either. In macro photography, internal focusing lenses can prevent the object from being touched accidentally with the front lens while focusing. As less weight has to be moved with internal focusing lenses, they are typically a lot faster.
Manual Focus vs. Autofocus
For manual focusing, camera lenses typically have a rubber focus ring installed on the lens mount that is mechanically linked to the moveable lens unit. For a very long time in the history of camera lenses, manual focus has been the only choice for photographers to control focus. In order to find the perfect lens position, a photographer had to turn the focus ring precisely while simultaneously looking through the viewfinder where the current focus situation could be viewed. Although considered to be less susceptible for errors and therefore still preferred by some photographers today, manual focus inevitably is time-consuming and cannot be used for fast moving objects. Admittedly, manual focus still achieves better results in dark environments.
Since the late 1970s, manual focus has increasingly been replaced by autofocus systems using electronic drive systems to change the focus lens position. A very common type of drive system consists of the ultrasonic motor that will be explained in the lens articles later. Also, some lenses have their focus mechanism driven my tiny stepper motors which are known for their high accuracy. Even those modern camera lenses with electronically driven focus lenses usually still allow manual focus if desired. In these cases, a toggle switch is installed on the lens case to choose between manual focus or autofocus. However, as far as complete autofocus systems in digital cameras are concerned, using electronic actuators to move the focus lens is only half of the story. If a camera should independently perform focusing, it has to know where the object is located and to evaluate the sharpness of this object. The following section describes a history on how digital cameras learned to analyze the sharpness of an object and which technology is used today.
Although the history of focus systems is not as long as the history of camera lenses, it is characterized by an evolution of several different technologies.
Coincide Image Rangefinder
The first focus aid systems used in photography were rangefinders. These systems relied on a principle called triangulation to determine the distance between the camera and the subject. Triangulation is a process where the position of a target (the subject to be photographed) is determined by looking at it from two different reference points located on a baseline. Two virtual lines are formed by the points and the target. The angles observed between the baseline and these virtual lines allow to calculate the distance between the baseline and the target. The following illustration shows the procedure.
The downside with this type of distance calculation is that two variable angles need to be measured. An easier calculation can be performed if the reference points (POV I and POV II) are positioned so that a right angle is formed with the target either in POV I or POV II. The arrangement and the resulting equation is shown in the figure below.
The camera was aimed at a subject so that it could be seen through the viewfinder. The viewfinder was typically located on an outer edge of the camera and included a semi-transparent mirror mounted with a fixed angle of 45 degree. A fully reflective, pivoting mirror was placed on the opposite side of the viewfider. A separate opening in the camera served to direct light towards the pivoting mirror which deflected it towards the semi-transparent mirror. The imaginary line from the viewfinder towards the target is called the sight line. The imaginary line that connects both mirrors is the base line. The sight line and the base line form right angles to each other. For the user who looked through the viewfinder, this setup displayed two virtually identical images of the same scene, superimposing each other. The image acquired through the viewfinder opening of the camera always was at a fixed location as the sight line always remains unchanged in position, but the image produced by the semi-transparent mirror – acquired through the rangefinder opening – was shifted horizontally depending on the target distance and the angle of the mirror. Only if the pivoting mirror was precisely adjusted by the user, both images in the viewfinder coincided, hence the name coincide image rangefinder.
The earliest of those rangefinders used in photography (around 1900) were so-called uncoupled rangefinders. This means that the rangefinder units were not connected to the lens and their readings had to be used to adjust the corresponding lens position by hand. These uncoupled rangefinders typically had a scale connected to the pivoting mirror that showed distance values. The camera lens in turn had distance values applied so that the photographer could directly set the lens to the right position according to the distance given by the rangefinder. The illustration below shows the principle on an uncoupled rangefinder.
In 1916, the first camera was introduced with a coupled rangefinder – the 3A Autographic Kodak Special. The coupled rangefinder was linked to the focusing system of a camera so that it could give an instant feedback to current position of the lens. This was achieved by a mechanical connection between the shift of the lens and the
rotation of the pivoting mirror. When focusing, the photographer changed the lens position and could directly see the superimposed images moving. Once the images coincided, the photographer knew the lens was properly focused and no further adjustment was required. Although this was a great invention, it was not until 1930 that this focus technology became widely used on cameras. The image below schematically shows a coupled rangefinder.
Coupled coincide image rangefinders have been in use until the 1970s. However, this technology is fully manual as it does not adjust the lenses independently. It was therefore a goal to create focusing systems that could detect the subject distance and set the lens to proper focus automatically. Therefore, these types of systems are called autofocus.
In general, autofocus systems can be divided into two types – active and passive. Active autofocus analyzes the target’s distance by the emission of a signal that is being reflected by the subject and is recaptured by special sensors on the camera. By contrast, a passive autofocus system only captures the available light in order to calculate the target’s distance from the camera.
This autofocus technology is separate from the optical system which means that it is not connected to the photo-taking lens or the viewfinder of the camera. Depending on the type of active autofocus, it is based on the emission of an optical or acoustical signal that is reflected by the target and can be analyzed by the camera.
- Sonar System
An active autofocus system that was introduced in 1972 with the Polaroid SX-70 camera is a sonar system – also referred to as ultrasonic autofocus. This time-of-flight system emits an ultrasonic chirp of approximately 60 kHz towards the subject. The chirp is reflected by the target and travels back to the camera. The camera translates the signal’s time-of-flight into a distance, and the lens position is set by an electronic actuator accordingly. The effective range of an ultrasonic autofocus system is roughly between 0.2 to 10 Meters. However, the problem with this type of measurement is that the result can deviate drastically if the chirp is unintentionally reflected by objects other than the target. If the target is located behind a glass plate, distance measurement by ultrasonic autofocus is completely impossible. Also, the size of the sonar system limits its use to larger cameras only.
- Infrared Triangulation
At the end of the 1970s, a different approach was taken to read the distance between the camera and the subject. The Canon AF35M – introduced in November 1979 – used infrared light to focus on its subject. This system is based on triangulation and can be thought of as an optoelectronic version of the standard rangefinder. Infrared triangulation works by projecting an infrared beam onto the target and detecting the angle of the reflected beam. The beam may originate from an infrared light emitting diode that is powered by the discharge of a small capacitor when the user presses the shutter button halfway. The resulting infrared flash may be a simple burst lasting just a millisecond or two, or it could be a series of ultrashort pulses. The returning beam is electronically filtered to make sure it matches the frequency of the emitted beam, thus avoiding a spurious signal. An actuator drives the lens to its in-focus position accordingly. In case that no infrared signal is detected by the sensor, the camera sets the lens to infinity focus. Still, a separate viewfinder is used so that the photographer can aim the camera at a desired target. The downside of this system is that it can be prone to errors, such as setting the lens to infinity focus if a particular target is not reflective for infrared light.
In summary, active autofocus systems have not been able to establish themselves due to low precision and their susceptibility to fault.
Contrary to autofocus based on signal emission, passive autofocus captures the available light and performs different calculations. The following descriptions analyze the individual approaches of passive autofocus systems and the technological developments over the last decades.
- Dual CCD Triangulation
This technology was announced in 1976. Researchers at the US company Honeywell International Inc. designed an autofocus system called Visitronic that became popular as the first passive autofocus system in the world. The pioneer camera that first applied the Visitronic system was the Konica C 35 AF camera that was released in 1977, even two years earlier than the infrared triangulation system. The distance between the camera and the target was calculated by triangulation and therefore, the technology can still be compared with the standard coincide image rangefinder. The only difference is that the evaluation of the superimposed images is done by the
camera electronics and not by the photographer.
With dual CCD triangulation, the camera acquires two images of the scene. One of these images comes from the target along a sight line established by the small aiming circle in the camera’s viewfinder. This primary image – called reference image – is formed on the surface of a linear CCD sensor array that produces a waveform that is the “signature” of the light distribution in the scene. Another identical sensor array receives the second image from the scene, producing a signature identical to the reference image if the second image is located in exactly the same position with respect to its CCD sensor array. This position on the second sensor is controlled by the angle of a pivoting mirror linked to the focusing system. The reference image is always formed on the same position of CCD I because it always arrives along the sight line – it is only the waveform that changes depending on the scene. The second image is formed on some portion of CCD II that depends on the target distance and focus settings. When the focus setting matches the target distance, the light distribution on CCD I and CCD II is exactly the same. The signals from the CCDs are fed to a comparator circuit designed to detect differences in waveform shape (light distribution) and phase (location). Any mismatch between signatures I and II causes the comparator to turn on a lens actuator through a control IC. The actuator’s direction is controlled by the plus/minus phase difference between the two signatures. When the two signatures coincide, the actuator stops.
Still, the pivoting mirror had to be linked to the focus system which often involved some very complicated mechanics. Another challenge at that time was to set the lens to its focused position automatically. The Visitronic autofocus system used a spring to drive the lens from an initial starting position and stopped the movement by an electromagnet.
- Solid State Triangulation
With the release of the Canon AF35ML in 1981, Canon introduced an improved version of the dual CCD triangulation technology. In their new detector, the pivoting mirror has been replaced by another fixed mirror and therefore the focus detection unit is completely free of moving parts which reduces vibration and noise. For that reason, this type of focus detection is called solid state triangulation (SST). Another improvement concerns the comparator algorithm which now instantly translates the phase difference between the two signatures into a digital signal. Finally, this autofocus system can read the current lens position and an adjustment can be made more efficiently. There can be slight variations to this technology, for instance, one longer sensor array can be employed instead of two linear sensor arrays. The figure below shows the principle of solid state triangulation.
An interesting combination was made by Canon where they integrated an SST autofocus detector into a lens case instead of the camera body. The Canon FD 35-70mm F4 was the first passive autofocus lens that was released later in 1981.