Battle of megapixels. Thoughts from Radozhiva. How many megapixels does your video surveillance need? How many megapixels is better?

The race for megapixels from digital photography has gradually moved into IP video surveillance. Our clients are increasingly asking for cameras of 3, 4, 5 megapixels and even higher. Most of them are absolutely sure that the higher the resolution, the more megapixels the camera has, the better it will show, the higher the detail of the frame will be. Manufacturers, to please consumers, produce high-resolution cameras; 12 MP IP cameras, in the now fashionable 4K format, are already being sold with all their might.

We decided to figure it out - does the video quality of IP cameras really increase with increasing megapixels? Is it worth paying extra for high-resolution cameras, NVR processing power, high network bandwidth and terabytes of disk space required for such high resolution? We selected several cameras from stock with different resolutions - from 1 to 5 megapixels. We also ordered several expensive 5 - 8 MP IP cameras from manufacturers for this test. This is who came to us for testing.

We gave preference to outdoor IP cameras with a fixed lens, because... they do not need to be adjusted, and errors in the tedious adjustment of varifocal lenses will not affect the quality of the video image. True, we did not find 5-megapixel cameras with a fixed lens and tested 5MP varifocal cameras. We installed all the cameras in the same place and pointed them at the opposite wall, where we have several homemade “test tables” hanging.

Let's see what we got. All frame shots were taken through the web interface of the cameras using the IE browser and the ability to save a still frame built into each camera. In the table below we have placed a reduced frame to a resolution of 640x480 (or 640 by 360 if the camera has a widescreen matrix with an aspect ratio of 16:9), as well as a crop (cut out of the frame) with a resolution of 200x360 pixels. It more clearly shows the quality of the “drawing” of small details of the image - in particular, the letters on the Sivtsev table (a table for testing vision).

To view a full-size frame from an IP camera, click on its small copy in the table.

What we noticed when comparing these frames:

1. Cameras have different aspect ratios. IP cameras with a resolution of 1, 2, 4 megapixels have a widescreen frame with a 16:9 ratio. And cameras with a resolution of 1.3, 3 and 5 megapixels are 4:3. Those. the latter have a greater vertical viewing angle. This is very important for those cameras that will “look” at an object from top to bottom. For such cameras, there will be fewer dead spots under the camera both near and far. It is interesting to note that a 3MP camera compared to a 4MP camera not only has a larger vertical viewing angle, but also a resolution: 1536 versus 1440 pixels.
2. Cameras have different viewing angles, and it depends not only on the lens, but also on the size of the matrix. Budget IP cameras with a 1/4 matrix and a standard 3.6mm lens have a horizontal viewing angle of no more than 60°. But the 5MP IPEYE camera with a 1/2.5 matrix has a wide viewing angle both vertically and horizontally (more than 110°). It’s true that the lens at its shortest focus has a distance of 2.8mm.
3. Well, the most important thing we wanted to pay close attention to is the resolution. If you look closely at all the frames, you will notice that, undoubtedly, as the resolution (megapixels) increases, the detail increases. But NOT PROPORTIONAL! Not huge. A 4MP camera does not improve the image by 2 times compared to a 2MP camera. Detail increases slightly. In any case, not a single camera could “cope” with the second line from the bottom of Sivtsev’s table. And already the 6th bottom line (the right letters “B K Y”) is confidently “read” by both cameras with a resolution of 4 and 2 MP.

Of course, here we need to make allowances for different viewing angles. After all, as the viewing angle increases, we seem to move away from the scene being filmed and the detail deteriorates. This is especially true for the 5-megapixel IPEYE camera - the viewing angle provided by this combination of matrix and lens is too wide. And if you make the angle on it the same as that of 2MP cameras (about 90°), then the letters of this table will be read more confidently.

It’s interesting that another 5MP IP camera with the same declared parameters (2.8-11 lens, 1/2.5 matrix) has a slightly narrower viewing angle at the shortest focal length than the IPEYE-3802VP. Detail is approximately at the same level, the picture is somewhat noisier in dark areas of the frame, although the cost of the BEWARD camera is several times higher. But it has a motorized lens and you can control the viewing angle while sitting in front of the computer. A picture with a maximum focus of 11 mm will then look like this:

Maybe someone needs this, considering that every time you change the focus of the lens, you need to either manually or by pressing the “autofocus” button to adjust the sharpness of the image. And this takes from 5 to 20 seconds. But here you can confidently read the second line from the bottom of the vision test table.

Subsequently, we tested a pair of 2-megapixel IP cameras with a 2.8 - 12mm varifocal lens, because... There is an opinion that they show better than “fixes”. Here's what we got:

As you can see, the result is not much different from the previous one. The detail is almost the same as that of 2MP IP cameras with a fixed lens. Even the expensive 2 megapixel (!) Hikvision camera (the retail price of which as of February 2016 was 21,990 rubles) with a viewing angle of 50 degrees set at the factory (and to change it, you had to open the camera, which we absolutely did not want) The readability of Sivtsev’s table was no higher than 5th line from the bottom.

Perhaps varifocal lenses have greater photosensitivity and IP cameras with them can “see” better in the dark, but this is the topic of a completely different test and another article, which we may turn to later. But varifocal lenses have virtually no effect on resolution. Moreover, the slightest inaccuracy in focusing settings can lead to disastrous results, and all megapixels will be useless. And anyone who has ever set up a varifocal lens on an IP camera will agree with me that this is oh so difficult, given the delay with which the signal from the camera arrives at the monitor.

This is the first camera with a 1/1.8 sensor size that we got our hands on. In addition, this camera is capable of delivering a stream at a speed of 25 fps with a 5-megapixel resolution (2592x1920 px). Others cannot do this yet. The maximum they are capable of is 12-15 fps at maximum resolution. The wide viewing angle of this camera is immediately noticeable. With a focus of 3.6 mm, it is wider than 5MP cameras with a 1/2.5 matrix with a focus of 2.8 mm. The resolution of the camera from BSP Security is at the level of other 5-megapixel cameras, even a little clearer. At least the contrast of the picture is higher. However, the situation is slightly overshadowed by blurring of the left side of the frame. Perhaps we were unlucky and came across a camera with a slight distortion of the matrix.

And finally, 4K IP cameras with 8MP resolution arrived at our warehouse. This is a hemisphere with a fixed lens DAHUA DH-IPC-HDW-4830EMP-AS. Here's a shot from that camera:

To open the frame in full resolution, right-click on the picture in your browser and select the “open image” menu item.

We didn't stop our test with office pictures; we also wanted to look at real footage of a street scene. To do this, we pointed our camera lenses at the nearest parking lot, visible from our window. We did this deliberately in rather difficult light conditions - early twilight. This is what we got.

Perhaps we chose a part of the day that was still too bright (17.10 - 18.00 in February), but all the cameras did an excellent job with such lighting. True, the 1.3 MP camera MT-CW960IP20 had a slightly darker picture than others, which is quite strange, because... the 1/3 matrix should have better light sensitivity compared to the 1/4 matrix.

As for the detail of the picture, the situation is similar to the results of testing in the office. Although it increases with increasing megapixels, it is not significant. Both 4 and 2 megapixel cameras were able to read Renault's license plate number. True last ones A little worse.

IP cameras with a resolution of 1.3, 4 and 5 megapixels with their wide viewing angles even “saw” the license plate of our van in which we carry all these IP cameras)). And the 5 MP camera even saw a car standing to the left of the van. The viewing angle is amazing!

In March, we received two more 5-megapixel IP cameras BEWARD and BSP Security for testing. Let's compare how they show on the street.

5 MP IP camera: resolution 2592x1944, matrix 1/2.5, zoom lens 2.8 - 11 mm
5 MP IP camera: BSP Security, resolution 2592*1920, matrix 1/1.8, lens 3.6 - 11 mm

The cameras were tested at the same time (18.00 in mid-March). It is interesting to note that despite the fact that the camera from BSP Security has a wider angle, it has slightly better detail. State The license plate on the blue Ford can almost be read, which is not the case in the BEWARD camera footage. The matrix size has an effect - 1/1.8 versus 1/2.5.

What conclusion can we draw?

1. The treacherous pursuit of megapixels is practically useless and only the manufacturers (well, let’s be honest, we, the sellers of these IP cameras, recorders and hard drives) benefit from more profit from them.
2. In the vast majority of cases, 1- and 2-megapixel IP cameras are sufficient. And if you need better detail of distant objects, then this problem should be solved not by mindlessly increasing megapixels, but by reducing the viewing angle using a varifocal lens. With this we will “bring the picture closer” to ourselves and will be able to see everything we need. And an increase in the number of video cameras. This solution may be a little more expensive, but it will definitely solve your problem. And perhaps the price of a pair of 2-megapixel cameras with a viewing angle of 50° (for example, “fixes” with a 6mm lens) will be less than the price of one 5- or even 4-megapixel with a viewing angle of 100°. But they will give us much more information about the observed territory.
3. It should be taken into account that increasing the number of pixels without increasing the physical size of the matrix only worsens the sensitivity of the video camera, because The pixel area becomes smaller and less light reaches its surface.
4. Real high-quality lenses with optics that allow you to get all the advantages of multi-megapixel matrices cost at least $1000. What can you expect from a 12-megapixel camera worth 20,000 rubles?
5. Well, the last thing to remember is that with an increase in “megapixel” you will additionally overpay for the processing power of the devices being recorded, storage devices (HDD), network bandwidth and traffic when viewing via the Internet.

P.S. We will continue to test IP cameras that come into our hands in this way. Several test samples have already been requested from various suppliers with resolutions ranging from 5 to 12 megapixels. Therefore, periodically visit this page to receive new information about the megapixel race in IP video surveillance.

P.P.S. If any of the manufacturers or suppliers would like to test their cameras on our “test bench” - welcome, contact us by e-mail: kb063_sobaka_yandex.ru

The term "megapixel" can be deciphered as one million pixels. That is, a 12-megapixel camera takes pictures that consist of 12 million tiny dots. The more these dots (pixels) in the image, the sharper it looks, the higher its resolution.

From this we can conclude that a camera with a large number of megapixels takes better pictures than one with fewer. But it is not so.

The problem is that nowadays they have more megapixels than they need. Let's think about screens: a FullHD TV has a resolution of 2.1 megapixels, and the latest 4K TV has a resolution of 8.3 megapixels. Considering that almost every modern smartphone has more than 10 megapixels in its camera, displays simply cannot display such a high resolution to its fullest potential.

It is unlikely that you will notice the difference between photos from modern cameras with different numbers of megapixels, since even the latest screens do not support such resolutions.

In fact, going over the 8.3 megapixel mark can be useful if you intend to crop your shots. In other words, if you take a photo with a 12-megapixel camera, you can cut off a significant portion of it. However, the image resolution may still remain higher than that of a 4K TV.

Advice. Don't go for cameras that have more than 12 megapixels. This amount is enough to spare, unless you are going to cut the pictures into fragments or edit them for professional purposes.

Pixel size is more important

An indicator that more accurately characterizes a smartphone camera is the pixel size. In the general list of characteristics, its numerical value is indicated in micrometers before the abbreviation µm. A smartphone camera with a pixel size of 1.4µm almost always takes better pictures than another with a pixel size of 1.0µm.

If you zoom in close enough to a photo, you can see individual pixels. The colors of these tiny dots are detected by microscopic light sensors inside the smartphone's camera.

These sensors are also called pixels because each sensor captures light for a corresponding pixel in the image. So if your camera has 12 megapixels, it has 12 million light-sensitive pixels.

Each sensor captures particles of light known as photons and uses them to determine the color and brightness of a pixel in an image. But photons are very active, and capturing them is not so easy. For example, instead of a blue particle, the sensor can catch a red one. As a result, instead of a pixel of one color, there will be a dot of another color in the image.

To avoid such inaccuracies, a light-sensitive pixel catches several photons at once, and special software uses them to calculate the correct hue and brightness of the point in the final photo. The larger the pixel area, the more photons it can capture, and the more accurate the colors in the final image.

Advice. Stop at cameras that have no more than 12 megapixels. A higher number forces the manufacturer to sacrifice pixel size in order to fit everything into a limited space. When comparing cameras with the same number of megapixels, choose the one with larger pixels.

Aperture

Another important camera characteristic that should not be neglected is the aperture. It is indicated using the symbol f divided by the numeric value. For example: f/2.0. Since f is divided by a number, the smaller it is, the better the aperture.

To understand the meaning of aperture, think about the size of a pixel. The larger it is, the more particles of light the camera captures, the more accurate the color rendition. Now imagine that a pixel is a bucket and photons are raindrops. It turns out that the wider the bucket (pixel), the more drops (photons) it gets.

The aperture resembles a funnel for this bucket. Its lower part is the same diameter as the bucket, but the upper part is much wider, which helps to collect even more drops. As the analogy suggests, a wide aperture allows the sensor to capture more light particles.

Of course, in reality there is no funnel. This effect is achieved through a lens that allows the camera to capture more light than its pixels can capture.

The main advantage of a wide aperture is that it makes the camera better at shooting in low light conditions.

When there is too little light, the light-sensitive pixels may not capture enough photons. But a wide aperture solves this problem by allowing access to more particles.

Advice. Remember, a smaller number means a wider aperture. So opt for cameras with f/2.2 or lower, especially if you often take photographs at night or indoors.

Image stabilization: EIS and OIS

Among other camera characteristics, you can find two types of image stabilization: optical - OIS (Optical Image Stabilization) and electronic - EIS (Electronic Image Stabilization).

When the camera sensor moves due to hand shake, OIS physically stabilizes the image. If you walk while shooting video, for example, each step usually changes the camera's position. But OIS keeps the sensor relatively stable, even if you shake your smartphone. As a result, the technology minimizes judder in videos and blur in pictures.

The presence of optical stabilization greatly increases the cost of the device and requires a lot of space for additional parts. Therefore, instead of it, electronic stabilization is often introduced into smartphones, which creates a similar effect.

EIS crops, stretches, and changes the perspective of the individual frames that make up a video. This happens in software and with the footage, so electronic stabilization can be applied even to videos recorded on cameras with OIS to make them even smoother.

By and large, having a camera with optical stabilization is better. After all, electronic processing of frames can reduce the quality and create a jelly effect on the video. In addition, EIS almost does not reduce the degree of blur in the pictures. But it is worth noting that electronic stabilization does not cease to develop, which confirms the quality of videos shot on devices.

Advice. If you can, choose devices with optical stabilization; if not, stick with electronic stabilization. Ignore devices that do not support either OIS or EIS.

We test flagship cameras “blindly” to make sure that the number of megapixels does not determine the quality of the image. We also understand the technical characteristics of cameras - what you really should pay attention to when choosing a new smartphone.

The buyer acts on the principle “The more megapixels, the better!”, without looking at other characteristics. And he buys a promoted stereotype, not high-quality equipment. Let's do an experiment. Here are 3 samples of photographs taken on flagship smartphones in 2015, and three typical photography scenes. Answer:

1. Which shot do you think is of better quality in each case?
2. How many megapixels do you think are in the camera it was shot on?

Flower macro photography

Shooting in low light conditions

Such tests are referred to as “blind assessments”. We deliberately did not indicate smartphone manufacturers so that brands would not bother our eyes. Well, how do you like the picture?

Pictures were taken on:

• HTC One (M9) - 20 MP;
• LG G4 - 16 MP;
• Samsung GALAXY S6 edge - 16 MP;
• Sony Xperia Z3+ - 20.7 MP.

Who is your leader in “blind evaluation”? Ours is Samsung GALAXY S6 edge. Please note: none of the smartphones coped with all three images perfectly. Because:

Conclusion 1

More megapixels does not improve the quality of photos. This is affected by a lot of other factors, including megapixel resolution.

Conclusion 2

It is extremely difficult to find the perfect smartphone for all shooting scenarios. Be prepared for the fact that a camera that takes stunningly detailed shots in daylight will fail evening tests or take poor macro shots, for example.

How to choose a smartphone with a good camera if the number of MP is not the main thing?

There are 4 key characteristics and a ton more additional ones.Remember!A smartphone with a good camera is chosen by:

• pixel/matrix size;
• aperture;
• image stabilization system;
• post-processing of images, camera’s own software.

What is all this anyway?

Pixels and matrix

The matrix of a smartphone camera is a mass of photosensitive cells. You press the shutter button, light enters the cells - gobbledygook! - the result is a photograph. The same number of megapixels does not mean the same number of cells. The same LG G4 and Samsung GALAXY S6 edge have 16 megapixels each, and the frame for both consists of 5312x2988 pixels (the models use a Sony sensor). And here is the frame on Huawei Mate 8 at 16 megapixels - from 4608x3456 pixels.

Camera matrices are of different sizes: LG G4 and Samsung GALAXY S6 edge have 1/2.6 inches, and Huawei Mate 8 has 1/2.8 inches. A smaller matrix means the size of the photosensitive cells is also smaller. Smaller cells receive less light: the light that hits the matrix quickly fills them, and the excess “spreads” over neighboring cells. Hence the inaccuracies in the rendering of details and “color spots”.

Flagships, traditionally, are smartphones with a powerful camera. The sensor size of the 12 MP camera of the iPhone 6s Plus is 1/3’’. IN Huawei Nexus 6P , Google's Android flagship, also has a built-in 12 MP camera, but with a 1/2.3’’ sensor. A smaller index after the fraction means a larger sensor size, which means, theoretically, better shooting. This is such a confusion :)

On a note: The camera phone Nokia Lumia 1020 - 41 megapixels and a 1/1.5″ matrix. This is almost the maximum for the sensor size in smartphones.

The larger the sensor, the better (the smaller the index after the fraction, the better).

Aperture

With aperture ( photosensitivity ) - exactly the same: the lower the index, the better. The f/х.y value shows how much light the camera can capture in a given period of time, how much the camera aperture can open to take a cool photo in low-light conditions. The maximum aperture today is f/1.7 (at Samsung GALAXY S7 and GALAXY S7 edge ) and f/1.8 (new flagship 2016 LG G5, LG G4, LG V10 smartphone, Xiaomi Mi 4 and Mi 4 LTE).

More often you can find models with f/2.0 ( Sony Xperia Z5) and f/2.2 (iPhone 6s Plus ), but in this segment the number of models exceeds a hundred.

The smaller the aperture index, the better.

The idea that new devices have stopped bringing innovative solutions to users is no longer new. Often all we get from a new model is an increase in megahertz, gigabytes and megapixels (and only occasionally something like a very useful heart rate sensor). Does the user need all this? Probably not always. Proof of this is with its QuadHD screen, the interface of which, according to reviews, slows down as a result, and this at a cost of 30,000 rubles. According to our colleagues from Techradar, the same goes for smartphone cameras.

The fact that comparing the quality of mobile photos by the number of megapixels is not objective has probably been conveyed to the majority. Nevertheless, when choosing a new flagship, many of us pay attention to the quality of the pictures it takes, expecting a proper increase. And yet, how many of these megapixels would be more than enough? According to the author of the article, Damien Demolder, in fact, only about 6 megapixels are needed for a photo of sufficient quality.

What are these conclusions made from? Firstly, from the practical application of your photographs. When was the last time you printed an A3 photo? Or even less, say, for an album? If landscape format is the most you'll ever print, then you don't even need 4 MP.

The science of mathematics, which is quite difficult to argue with, states that for the best image quality you need 300 pixels per linear inch, that is, 3000 x 2400 pixels for a 10 x 8 inch format, which is approximately comparable to the A4 format. But then again, when was the last time you printed your photos?

If your footage, like most of us, is screen-only, then you won't need more than 8MP, and that's for viewers with larger 4K screens.

From the point of view of a professional photographer, it would be rational to calculate which format he has to work with most often and, based on this, choose a camera with the appropriate resolution. But the beauty of mobile photography is that it is not a professional activity, but your personal entertainment. And if you want your smartphone camera to have 100 megapixels, then who has the right to stop you from buying such a device if it is on the market? As they say, whatever the child enjoys.

On my own behalf, I would like to note that HTC clearly demonstrated how to fail a flagship on the market by not giving it a good camera. Four megapixels, which are also “Ultra”, turned out to be clearly not enough. Personally, I think that more is not less, because who doesn’t love high-quality photographs? And the quality of mobile photos is certainly improving. If for this the manufacturer needs to sell me more megapixels, so be it.

For many, smartphones are already replacing point-and-shoot cameras on vacation, and that’s great. I would like development in this regard to not stop. You see, one day, basically filled only with mobile photos, will delight its subscribers with shots of the starry sky. It would be great.

If you understand photography and a little about smartphones, then we congratulate you and this article will be more of a reminder for you. If you are, as they say, a “newbie” in this matter, then read on.

What is a megapixel? Simply put, a megapixel is one million pixels. Therefore, if you know the resolution of an image, you can calculate how many megapixels it has. Many modern smartphones have a screen resolution of 1080 by 1920 pixels. Thus, a photo of sufficient clarity will have 1920 x 1080 = 2,073,600 pixels or 2 megapixels. So for normal video communication, 2 MP webcams are more than enough.

Taking this simple information into account, think about whether you need a huge number of megapixels in your camera? The example above shows that many megapixels do not matter if you just need to post photos to Instagram or other social networks. The number of megapixels, taken separately from other camera characteristics, is not a guarantee of high-quality photos. In fact, megapixels only matter in two cases.

1. If you need to print the photo.

As you know, photo printing has significantly lost its popularity with the advent of digital photography. Most people are quite happy uploading photos to the Internet or storing digital copies of them on flash drives and cloud storage. Almost no one thinks about printing anymore. If you still need to print a successful photo, then it’s worth thinking about megapixels. The more pixels in the original image, the sharper the image will be when printed on photo paper.

This table shows what the quality of printed photos will be, based on the ratio of the number of megapixels of the camera and the size of the photos.

When it comes to everyday needs like printing a memorable photo, you don't need more than 7 megapixels. Even if the size of the printed photo is 30x40 cm, then due to the relatively large viewing distance you will not notice any shortcomings. You won't look at the gloss of photo paper, will you? Framed and on the wall.

2. If you need to crop a photo

Everyone seems to have already learned that when taking photographs you should not use digital zoom. Otherwise, no matter how high-quality the camera is or how colorful the picture is, the wrong zoom will turn it into an unpleasant grainy something, leaving you dissatisfied with both the camera and your photography ability. Another thing is when your camera has a sufficient number of megapixels and you can easily crop a finished photo without any concerns about loss of quality.

Fortunately, cameras in modern smartphones already sport 12 megapixel or higher cameras by default. Although, even with some 4 megapixel camera, you can crop your photo in half and put it on your computer desktop without fear of degrading the quality. The truth is that, for example, a camera with “only” 20 megapixels will take a picture many times better than a 41 megapixel monster with mediocre characteristics.