Digital Camera Anatomy: What's Inside It?
The main principle of the camera’s operation has not changed since its invention - light rays reflected from photographed objects are focused by the camera’s lens and create a reduced image on a flat surface located a short distance from the lens. But if in the case of a traditional camera a piece of film turns out to be this surface, then in a digital camera the rays fall on the sensor, the task of which is to convert the photon flux (and these, in fact, these light rays) into an electron flux, that is, simply, into an electric one current. Then the current will be amplified, converted into a set of bits, processed, and finally written to the camera’s memory. The sensor is divided into so-called pixels — cells several microns in size, each of which is responsible for registering one image point. Traditionally, pixels have the shape of a square and are arranged in “rows” and “columns”, so the sensor is often called a “matrix”.
The main material for the manufacture of the sensor is the cornerstone of modern civilization, hiding inside a wide variety of objects, from consumer electronics chips to busts of Hollywood stars - silicon (aka silicone). Photons incident on the surface of the sensor knock electrons from the outer orbit of silicon atoms, forming an electron-hole pair. The fate of the electron that suddenly found freedom (its path to the amplifier) depends on which of the two common types of sensors (CCD or CMOS) currently used in the camera.
CCD (ChargeCoupled Device, aka charge-coupled device, CCD) is currently the most common type of sensor. Like competing technology (CMOS), it was developed by Bell Laboratories in the late 60s and was originally intended for use as computer memory. However, in the early 70s, the first commercial CCD matrices appeared with a resolution suitable for use in television cameras. The principle of operation of the CCD is based on the accumulation of electrons released as a result of the photoelectric process, directly in the pixels of the matrix (in the so-called “potential wells”), then the line-by-line movement of the charges accumulated in these wells, to adjacent lines (hence the name of the technology), and then - and to the edge of the matrix. The line that appears on the edge is likewise pixel-by-pixel shifted to one of the angles where the charges enter the amplifier input and are converted into electric current. How are the "potential wells" arranged and how does the charge shift from the well to the well? A thin layer of silicon oxide, which serves as a dielectric, is deposited on the surface of the silicon wafer of the sensor, and behind it is a metal layer (electrode), onto which a positive potential is applied during the matrix “exposure”. As a result, an electric field arises in the adjacent semiconductor, which repels holes and attracts free electrons. The longer the matrix was exposed and the more photons hit a single pixel of the sensor, the more electrons accumulate in the potential well of this pixel. If a larger potential is applied to the electrode of the adjacent row of the matrix, the charge will move to the neighboring, “deeper” potential well. Thus, during readout, charges are transferred to the edge of the matrix and towards the amplifier. The amplified signal is fed to the input of an analog-to-digital converter (ADC), and then it is subjected to digital processing.
Another growing technology for manufacturing digital camera sensors is CMOS (Complimentary Metal-Oxide-Semiconductor, or Complementary Metal-Oxide-Semiconductor [circuit], CMOS). Unlike CCD, CMOS converts the charge to voltage (i.e., gain) directly in the pixel itself, and then provides direct access to the contents of an arbitrary pixel in the same way as it does in computer memory. Random access to image elements allows you to increase the speed of the camera during pre-reading, and due to the use of technology for the manufacture of the sensor, widely used in the manufacture of many digital microcircuits, CMOS sensors can be integrated with other components of the digital camera - the ADC and even the processor that performs image processing. The advantages of CMOS include significantly more modest power consumption, as well as the compactness and low cost of the entire camera design, because the amplifier and ADC as separate components are no longer needed. However, until recently, CMOS was used only in the cheapest cameras, since it could not compete with CCD in image quality: due to the fact that each pixel uses its own amplifier, a fair amount of discrepancy was noticeable in the resulting picture. Yes, and with the sensitivity of the matrix there were problems, because due to the abundance of electronics on the matrix there was not so much space for the actual photosensitive elements. However, the increased quality of silicon wafer production, an improved amplifier circuit and advanced noise reduction technologies now allow CMOS sensors to compete quite well in quality with CCD. So even Canon a few years ago ventured to produce professional-level digital "DSLRs" (D-30, now D-60 and D-10) based on CMOS sensors.
It is easy to imagine that the greater the distance from the center of the frame, the smaller the angle at which the rays of light fall on the surface of the sensor, and at a right angle, the rays fall on the sensor only in the center of the frame. Conventional film is not particularly sensitive to the angle of incidence of light, but in the case of a sensor, this is critical. So the manufacturers of some matrices also have to go to hefty tricks, placing a microscopic lens over each of the pixels, focusing a ray of light in the right place and at the right angle.
Any of the above sensors in itself is a monochrome device. In order to make the matrix sensitive to color, tiny filters from “primary” colors are applied to the pixel surface. The most common Bayer pattern, in which staggered green pixels alternate with blue and red (see. Fig.). The fact that green pixels are twice as large as each of the other two colors is explained by the increased sensitivity of the human eye to green, as well as the fact that green most significantly affects the subjective sharpness of the picture. Thus, in a 3 megapixel camera, the matrix has 1.5 million green pixels and 750 thousand blue and red pixels each. How does the camera manage to provide an output image in which 3 million full-color pixels? The camera processor uses interpolation in order to calculate the missing color information for each point in the image from the intensity of neighboring pixels of a different color.
Some cameras are able to produce pictures with a resolution that exceeds the actual resolution of the sensor. Where the extra pixels come from, you probably have already begun to guess. As with colors, they are obtained as a result of interpolation. Many, not without reason, believe that this approach does not increase the quality of the image, but the file size. After all, additional information from this does not appear, and you can increase the number of pixels in the picture yourself, with your favorite graphics editor.
Another interesting ambiguity that you can often notice in the technical specifications of the camera is a slight discrepancy between the “nominal” and “working” number of sensor pixels. As a rule, the difference does not exceed 5%, so it is hardly significant in practice, but often does not give rest to curiosity. There are several reasons for the fact that not all pixels of the matrix are working. First, microscopic defects are inevitable in the manufacture of the sensor. In addition, the camera uses part of the matrix space for official purposes — for example, to determine the magnitude of “dark currents”.
It is not that simple
It is generally accepted that the quality of the “digital” image depends primarily on the number of pixels on the sensor. It is this parameter that the manufacturer proudly writes on the front panel of the camera, and in various reviews and comparisons this figure is considered to be the determining class of the camera. In part, this opinion is justified, because the more pixels the frame is divided, the more details it conveys and the more sharply the image looks. However, simply “slicing” the sensor into a larger number of pixels is by no means a solution to the quality problem, because with the same matrix size, the more pixels there are on it, the smaller each of them. And with a decrease in the size of the pixel, its sensitivity also decreases, since less light falls on it. Thus, the signal will have to be amplified even more, and when amplified along with the useful signal, harmful noises arising in the matrix for many reasons will also be amplified - these are the so-called “dark currents", that is, the charge removed from the matrix even in the absence of light, and thermal noise from microcircuits warming up during the camera’s operation, and electrons flying into the neighboring pixel during exposure or “lost” when reading. Therefore, one of the most significant formal indicators determining the quality of a digital camera is the dynamic range, expressed in decibels and calculated as 20 x log10 x (maximum signal level / noise level). Certain conclusions can be drawn from the size of the matrix. For example, the image quality of the already mentioned 3-megapixel Canon D-30 is noticeably superior to many modern 5-megapixel soap dishes, and this despite the CMOS sensor! (In fairness, it should be noted that the D-30 is not cheap.)
Having discussed in so much detail the "heart" of the camera, its sensor, let's now move on to the "brain", that is, to the processor. Its role is to make a beautiful picture out of information about the intensity of individual pixels emerging from the analog-to-digital converter. First of all, for this it is necessary to restore color information and, in some cases, increase the resolution of the image due to interpolation. Further processing may include correction of white balance, brightness and contrast, as well as various visual effects - for example, image tinting or even correction of optical defects due to software sharpening. The final stage of processing is the compression of the picture - of course, so that more pictures fit into the camera’s memory. The speed of the camera directly depends on the speed of the camera’s processor and the amount of buffer memory, that is, how quickly you can shoot a series of shots and how many frames the camera can take before you think hard.
Pixel Per Pixel
As already mentioned, it is believed that the number of megapixels in the camera determines its class, and after it the price category. Therefore, when buying, you may be puzzled by the question: “How many megapixels are needed for happiness?” The answer to this question, of course, depends on what you are going to do with the pictures. If your task is to simply upload them to the web or send them to friends via email, you will probably find even a megapixel camera suitable. You still can’t make out a lot on the monitor, and a million pixels is about as much as usually fits on a computer screen. However, the familiar process of viewing paper photographs has some special appeal, so not everyone is ready to exchange it for clicking with the mouse in the browser. And, most likely, sooner or later you will want to give a tangible form to your digital creations. Decent print quality implies a resolution of 300 dpi, so for regular 10x15 prints you will need an image that is already almost 1800x1200 in size, i.e. about 2 megapixels. And if we take into account the ability to frame the image and various errors during subsequent computer processing, then 3 megapixels seem more appropriate. If your thoughts are warming about the possibility of hanging a large-format print on the wall, then you should think about buying a camera with a higher resolution, and with it, and whether the convenience of digital processing will pay off the difference in price between a decent "digital camera" and good film camera.
With the increase in the quality of the sensor and the approach of its resolution to the film, the lens becomes the most important component of the camera to get a good photograph. Fortunately, the relatively high price of digital cameras allows most manufacturers not to save on optics. In addition, the frame format of a digital camera is usually much smaller than that of a film camera, so optics also require more modest sizes, and therefore cheaper. Therefore, compact digital cameras often get lenses with decent quality and good aperture. Many eminent manufacturers of electronics that do not have their own experience in developing lenses produce "digital cameras" with the optics of well-known companies. For example, Panasonic mounts lenses from Leica, Sony - from Carl Zeiss, Fuji - from Nikon, and Casio - from Canon. One of the main parameters of the lens is the focal length: the angle of view and the magnification of the lens depend on it. With the light hand of Leitz, for more than half a century, most photographers have been shooting with 35 mm cameras and have long been accustomed to the focal lengths of lenses designed for 24x35 frame format. So, lenses with a focal length of 50 mm have a viewing angle like that of the human eye. 28–35 mm - classic wide-angle lenses, convenient for shooting landscapes, as well as mounted on most “soap dishes”. 85–135 mm - telephoto lenses most suitable for portraits. 300-500 mm - televisions commonly used for remote shooting of football, wild animals and important people. As already mentioned, the frame format of digital cameras is much smaller, so the focal lengths appear there are completely different. But in order not to create confusion, manufacturers often indicate an analogue of the focal length for a 35 mm frame. Например, настоящее фокусное расстояние зума у Minolta Dimage 7 — от 7, 2 до 50, 8 мм, а аналогичный объектив для 35 мм имел бы фокусное расстояние от 28 до 200 мм (то есть, по сравнению со стандартным объективом 50 мм, он обеспечивает четырехкратное увеличение и почти двукратное уменьшение изображения). Многие изготовители встраивают в камеру функцию «цифрового зума» — попросту, возможность взять кусок изображения из центра матрицы и «растянуть» его до размера всего кадра в процессе цифровой обработки. Как и в случае интерполяционного увеличения разрешения камеры, практическая полезность такой функциональности весьма невелика, ведь любой графический редактор справится с этим ничуть не хуже камеры. Профессиональные цифровые камеры допускают установку сменных объективов со своих пленочных аналогов. Однако сенсоры с размером полноценного пленочного кадра (24x35 мм) появились лишь недавно, да и стоят ощутимо дорого даже для профессиональной техники. До недавнего времени в большинство зеркалок устанавливали матрицы размером 15x22 мм, так что фокусное расстояние обычных объективов автоматически увеличивалось в 1, 6 раза. Что, с одной стороны, даже неплохо, потому как делало более доступными дальнобойные телевики, но, с другой стороны, практически лишало фотографов «сверхширокоугольных» объективов.
Пленка не сдается
Несмотря на многочисленные плюсы цифровой фотографии, пленка все еще не окончательно сдала свои позиции. Скорострельность и время реакции на спуск затвора даже у профессиональных цифровых камер не могут сравняться с многими пленочными моделями любительского уровня. В некоторых условиях старые механические камеры оказываются единственным решением, так как не требуют зарядки. А если вы едете в путешествие с цифровой камерой, вам придется подумать не только о том, где и от чего ее заряжать, но и о ноутбуке или хорошем запасе недешевых цифровых носителей, чтобы было где складировать отснятые кадры. Если вы продвинутый фотолюбитель, то цифровая техника обеспечивает для вас далеко не лучшее отношение «цена — качество». Покупая даже хорошую цифровую камеру (около $1000), вы вынуждены довольствоваться несменным объективом и серьезными неудобствами ручной фокусировки. В то время как за те же деньги могли бы купить весьма серьезную пленочную «зеркалку» с парой неплохих объективов.
И все же рано или поздно «цифра» победит — победят оперативность получения снимков, отсутствие затрат на пленку и неудобств, связанных с ее проявкой, компактность и надежность камеры (ведь можно обойтись без механического затвора и прыгающего зеркала). А главное — победит возможность обрабатывать и печатать свои снимки самому, без всех неудобств, связанных с фотохимическим процессом, и не прибегая для этого к помощи «минилабов».The article was published in the journal Popular Mechanics (No. 6, June 2003).