We test monitors in a few different ways. We test most models using an older MicroVision SS220 semi-automatic colorimeter. This robot doesn't only measure colour quality and temperature, brightness, contrast and gamma, but also brightness under horizontal and vertical viewing angles of 45 degrees. It also measures the uniformity (brightness distribution) with a brightness comparison on 25 points.
Because the accuracy of the SS220 colour fidelity test is limited to the older CIE1976 standard, we also use a X-Rite i1 Display Pro colorimeter combined with advanced Spectracal Calman 5 software to test the monitors. We record maximum and minimum brightness, contrast, gamma values and colour- and gray deviation based on CIE1994. We do this because the modern CIE2000 standard still isn't widely supported yet and because we have a lot of comparing possibilities with this older standard. We will introduce the new one step by step into our test procedure. For measuring the monitors out-of-the-box settings, we use the i1 / Calman combination standard; if there is an sRGB-modus, we adjust the screen to this for this test. If there is an AdobeRGB-modus, we test that separately.
In addition to the mentioned display measurements, we normally also measure response speed, overshoot and undershoot and the input lag. We always do this with gaming screens, with other monitors only when this provides us with meaningful results. For the first three tests we use an photometer combined with an oscilloscope. For the input lag test we use a visual comparison to a CRT monitor utilizing high speed photography, as well as (if possible) a Leo Bodnar input lag tester. This tester is limited to 1920x1080 regarding the output signal, and our experience is that results for screens with higher resolution are not reproducable and/or unpredictable. That's why we don't always mention those results regarding high resolution monitors.
Finally, we measure power consumption using calibrated power gauges, measuring a completely black screen and a completely white screen, as well as measuring the screen in stand-by modus and when the screen is turned off.
Settings for tests
All tests are conducted with out-of-the-box settings: most users will use those. We make a few exceptions: for colour temperature measurements we we try to adjust the screen as close to 6500 Kelvin as possible, for sRGB and AdobeRGB measurements we use as said the corresponding settings, if those settings exist on the monitor. If there is no sRGB setting, we apply Calman-measurements using standard settings. Furthermore we conduct all tests at 100 percent brightness, to get an honest comparison and to keep tests manageable.
Readers often ask us to calibrate monitors, or to adjust settings in such a way that colour, brightness and contrast are optimized. We do not do this however, because of two reasons. Firstly, most monitors and almost all monitors for private use can only be calibrated using software. This means you set up a colour profile for your monitor and graphics cards as a combination: this profile only works with that specific combination, and only for your specific screen. Even identical monitors can have small differences between them, in need of a different profile. In this way you actually adjust your graphics card's signal.
Only monitors for professional use and specifically the ones used for graphic applications have a hardware-based calibration possibility. Still, it remains true that the settings we use aren't necessarily sensible for a monitor the reader buys. The second reason is calibration being a very time-consuming process. What we can do is mention if our test results imply that calibration would lead to an (almost) perfect image. Calibration always leads to a better image, but the necessary colorimeter is too expensive for most users to do this themselves - good ones cost more than most monitors for private use.
Regarding brightness, we prefer scores above 250 cd/m² for maximum brightness and under 0,3 cd/m² for minimum brightness. Brightness higher than 300 cd/m² is factually not useful, except for extremely light environments, so this doesn't influence any scores. Maximum contrast ideally ascends 800:1, 1000:1 is good, above this would be outstanding. What should be mentioned is that contrast values in non-darkened rooms usually come out much lower, and that contrast values above 100-300:1 are exceptional in reality. The scores are mainly performance indications: the higher, the better.
Regarding colour temperature, a score as close to 6500 Kelvin is desired; this corresponds to normal daylight circumstances. Scores between 6000 and 7000K are good enough for average use, values out of this range result in worse test scores. Values between 6400 and 6600K are outstanding.
Colour- and grayscale deviation are acceptable when they are lower than 5, good when they are lower than 3 and outstanding when they are lower than 2. We will not recommend screens with values above 5 for anything more often than incidental image editing.
We record reaction times in parts (rise & fall) as well as combined. The most important values aren't necessarily the combined 0%-100%-0% and 20%-80%-20% values, but the 'optimal' measured values. They represent the results of those settings which result in the least over- and undershoot, but does come close to the minimal desired speed of 16 ms. We record the results without overdrive as well as with maximum and optimal overdrive, if there are corresponding settings. If there aren't too many artefacts due to under- and overshoot, values of 16ms and lower are acceptable, 10ms and lower are good and 6ms and lower are outstanding.
Input lag stays a phenomenon about which not everyone agrees if it's actually troublesome, but values above 16ms result in us not recommending a screen for gaming. It is important to remember that results of the CRT comparison have to be read excluding the panel's response time and the Leo Bodnar tester's including those response times.
For the power consumption, the verdict about the measured values depend on the size of the screen, the resolution and the colour space model (AdobeRGB monitors consume more because of a more complex backlight). However, it is a fact that consumption in stand-by above 0,5W and consumption when turned off above 0W aren't desirable. Moreover, stand-by power consumption above 0,5W conflicts with EU laws.