What is discretization, any professional in the field of digital photography knows today. However, many people who are just starting to get acquainted with this area do not know its basic features, as a result of which they can make certain mistakes.
What is it?
What is discretization? This is an undesirable effect, which leads to the fact that the quality of the photo noticeably deteriorates. This phenomenon can be associated with any device or a process in which information is divided into several separate samples. In this case, discretization can be considered as a type of interference pattern if there is a definite correlation between the frequency of a given phenomenon and a certain periodic structure in the data.
The human eye is constantly striving to perceive a certain ratio as an interference picture, which can obscure the real meaning of an image. Considering the examples of what discretization is, it is worth highlighting the moire, which is not quite its exact effect, but it can show how a person is misled when two patterns begin to interact with each other, forming a third.
What is moire?
Moire is an incomprehensible wave-like pattern that was not originally present at the subject. This effect most often occurs on various images that are obtained using digital devices. And the problem here is that the pattern of the object is superimposed on the pattern of placing pixels on the matrix, as a result of which a third appears, which is called moire.
In the predominant majority of cases, this effect occurs on detailed high-contrast images that do not correspond to the initially configured resolution of the sensors. In particular, it can often be found if objects such as hair or fabrics are removed, as well as scenes that contain a large number of repetitive details. Often moire cannot be found in nature, as it occurs on images that are obtained using a digital camera or subsequently incorrect scanning.
Quite often, in modern digital cameras, in order to reduce this effect, a specialized optical low-pass filter is used, so if you are really going to professionally engage in photography, then you should definitely think about a model with this function that can ensure the proper quality outside depending on secondary conditions.
Discretization in modern cameras
In modern digital devices, the discretization effect is caused by the fact that information is divided into several samples with a regular interval. In particular, one of the patterns in this case will be the arrangement of pixels on the matrix, the second pattern will be any elements in the image that can repeat over a large area or that change through a certain number of pixels in the transverse or longitudinal direction.
For those who do not understand what discretization is and when it occurs, a concrete example can be given. When there are not enough pixels in order to transmit reliable information from the image, then in this case it is not in the best quality. In the standard version, it would be enough just to choose a higher resolution, thus guaranteeing the right number of pixels to transmit details not in the picture with a certain accuracy, but if the number of pixels were insufficient, then we could just see a small number of picture elements.
However, in reality this is not entirely true. The sampling theory says that in reality the situation is much more negative, and if we do not have enough pixels to take a particular picture, then the image quality will constantly deteriorate.
How many pixels do you need?
Take, for example, a situation where there are simply 20 black and white lines in the image, each of which has a width of 5 pixels. In this case, if at least one pixel is provided on each line, then a picture can be recorded. Naturally, if the pixels do not fall clearly in the center of each line, then in this case each pixel will turn out to be gray, not white or black, and its hue will directly depend on how the pixel is located relative to the lines.
If the number of pixels in the picture decreases, then in this case some of them will start to appear between the lines, as a result of which the above pattern appears on the image, which will constantly change depending on the ratio between the lines between the lines and the number of pixels. Of course, such an image will no longer be an exact reproduction of the original, since the regular structure of the lines will already be noticeably disturbed. It is this phenomenon that is commonly accepted in professional circles to be called "data discretization."
What to do?
To solve this problem, you need to soften the image a bit before reducing the number of pixels. Thus, you can completely get rid of sharp edges on each line, and pixels will be able to take intermediate values. In other words, the image becomes softer, but the overall impression of the image is preserved.
How does this affect the image?
Of course, repeated and regular line structures are rarely seen in photographs of various natural objects - their presence is often limited to photographs of various artificial structures, such as buildings and so on. However, in any case, the sampling depth can be impressive, so this effect should always be avoided when shooting any objects.
It should be noted that the quality of images can be completely different even if they have the same number of pixels. Indeed, among other things, the difference between the images can also lie in the way in which they were obtained. For example, in one case, the image can be slightly softened by passing it through a low-pass filter to obtain intermediate pixel values before reducing the size, while the other image can simply be reduced in size without making any additional changes to it and without getting intermediate values at the boundaries of objects where too sharp changes in brightness are observed.
How to check?
To understand how this works, just take one shot and then make a copy of it. Filter the original when resizing using the so-called bicubic filtering option, which is available in standard Adobe Photoshop. Thus, the picture will be softened. Despite a noticeable decrease in pixels, the transitions between tonalities ultimately turn out to be smooth and quite suitable for the available number of pixels.
After that, we reduce the previously made copy of the image, say, to 30% of the original size, using the option "nearest neighboring point" in the same program. It is this operation that will ultimately give you the discretization effect that will be there.
Sound sampling
Sound sampling is filtering the sound wave before it is saved to the sound file. In other words, the final file will not be an exact copy of the sound wave, but only an approximate one. On the one hand, sound sampling provides a certain reduction in the volume of the saved file, but on the other hand there is a mass of sound vibrations that do not have to be stored on the hard drive.
This sound filtering is called “sampling rate”. It is worth noting that, in fact, only in nature is there sound without this effect, although few know about it. The sampling rate is the imposition of a specific grid on a sound wave, as well as the recording of only certain key elements.
It would be quite difficult to record completely the entire sound wave. For this reason, it is much more common to encounter a situation where two-channel sound recording with a sampling frequency of 44.1 kHz was performed. The latter is chosen most often, since this is the most optimal parameter.
In principle, when considering sound processing, you need to pay special attention to such parameters as the coding depth and sampling rate, because the higher these indicators, the more the digital signal will correspond to the analog one.