Testing cameras

Determining the actual characteristics of all types of video cameras

At the dawn of digital video surveillance, every TV line mattered, and everyone was interested in video camera testing. Nowadays, you can hardly find anyone using a resolution test chart anymore. And that is a mistake.

Yes, modern cameras now have a huge number of pixels. But because of this, you may be losing a lot without even realizing how much performance the IP camera you are buying actually delivers. The problem is not only that you may be sold a 2-megapixel camera at the price of a 5-megapixel one, but also how much you later pay to process those meaningless extra 3 megapixels.

Analyzing such inflated resolutions requires, at the very least, an additional physical CPU core; recording demands several times more disk space; network transmission requires significantly wider bandwidth; displaying the image on a monitor takes more screen space… And this is only the technical side of the issue. The main problem is that the project itself will be implemented incorrectly, and the calculated megapixel performance at the camera installation site will fail to provide the expected level of detail. You will not see the faces, license plates, or other identifying features you expected during the design stage. And then you end up climbing the pole again to replace the camera, repeating the installation work — wouldn’t it be easier to test everything properly from the start?

As an example from our tests 

The conclusion is disappointing: the actual resolution of all cameras turned out to be 1.5–2 times lower than the declared specifications.

Yes, this is not catastrophic in itself — you are not overpaying that much for the cameras, especially since cameras are constantly becoming cheaper. However, computing resources are not getting cheaper, which is why it makes sense to adjust the camera settings in order to save all the other system resources. For example, instead of using a resolution of 2592 × 1954 on a “Zodikam” camera, set something closer to 2200 × 1350, and your hardware will breathe much easier. In both cases, the image quality will remain practically the same, while resource consumption will be reduced by more than half. For example, you could connect twice as many identical cameras to the same computer. And that is no longer a minor saving.

Installation work — especially mounting a camera on a pole using a lift platform — is also far from cheap. You install a 2-megapixel device expecting it to capture license plates, but it fails to see them clearly. Next time you buy a 3-megapixel camera, and it still does not work as expected. Test everything on the ground before installation!

And it is not only about the pixels — the test “snapshots” clearly show that along the edges of many cameras, and in some cases even starting from the center, you will get complete image degradation.