We have based our test method around an X-Rite i1 Display Pro colorimeter and an X-Rite i1 Basic Pro spectrophotometer, which we can use to calibrate the colorimeter for every monitor test. Using our self-made accessory (long live 3D printers) we can use this colorimeter to measure the viewing angles. For the analysis of the data we use a new workflow developed in SpectraCal Calman 5.
We test monitors on a great number of aspects. We measure them 'out of the box', but if they have an sRGB-mode we test this separately. This is also the case with an AdobeRGB-mode. An exception to this are the monitors that are clearly adjusted for a wider colour space out of the box – in this case we measure using a corresponding colour space. The standard one we use is the sRGB-triangle.
We use our colorimeter to set the monitors to a brightness of 150 cd/m² (or a value that comes as close to this as possible). This brightness is representative for what you will use in most scenarios in a normally lit room, and can be displayed by practically every monitor. By configuring all monitors to the same brightness of white, we create an equal playing field. Of course we still measure the maximum brightness, as well as the corresponding contrast and power consumption. Below we will briefly discuss the different segments of the test.
Brightness and contrast
We measure brightness and contrast in both 150 cd/m² and the maximum brightness, as well as the black values in both settings. Aside from that we measure the contrast between a white and black surface, surrounded by a 50% grey surface (transvers measurement), the checkerboard contrast and the maximum contrast between a fully white and fully black surface.
For the uniformity we first look at the ratio between the lowest brightness in a white area compared with the highest measured brightness, expressed as a percentage. For example: the least bright part of the screen was 77% of the brightness of the brightest part. We also measure the ratio between the average of a measurement of 15 points and the highest brightness. Aside from that we perform the uniformity measurement with a black surface as well, showing severe clouding and backlight bleeding – if it is present. All uniformity measurements result in five screenshots per tested monitor: the uniformity of black, white, the contrast ratio at all (15) points, the colour temperature uniformity and the relative colour differences in comparison with the middle, expressed as DeltaE-value.
While we only measured the remaining brightness at an angle of 45 degrees horizontally and 45 degrees vertically above and below, we expand on this now. We still measure at an angle of 45 degrees, but aside from the remaining brightness we also measure the colour changes in comparison with the middle, expressed as DeltaE-value. This is based on the average of red, green, blue, cyan, magenta and yellow, plus 100% and 75% white. Because we also calculate the standard deviation of these measurements, we can also give an indication of whether or not there is a colour shift: a high sd means that one or more measurements deviates a lot resulting in a colour shift.
Aside from the average colour deviation we also use a screenshot to show the deviation of white relative to the middle from each of the measured corners. We do this for every tested product as this allows any colour shift to become clear.
For some time now we have been performing colour measurements using CIE1994 as well as CIE2000. With our new test method CIE2000 becomes the standard. This result is now based on a larger amount of measurements, the so-called ‘Colour checker’ where we take a look at a lot of common hues. Once again we show the standard deviation as well as the average deviation, to give an impression of the amount of fluctuation of our measurements. A smaller number means there is a more constant degree of deviation. This is also the case for our grayscale measurements, that we used to do in ten steps but now perform in twenty; this will result in more accurate data. With every tested product we will also add a screenshot of a so-called saturation sweep in order to show the deviation of the base- and support colours in the entire colour space. Of course we will also publish the screenshots of the colour measurement and colour checker results.
We have added measurements of a white and black screen at 150 cd/m² to the existing measurements at 100% brightness (white, black, standby and off). This will show a more realistic image of the actual power consumption of the tested monitors.
Something that has not changed (yet), are our tests for gaming purposes: reaction times and input lag. For this we still use a photo sensor with an oscilloscope and a combination of a Leo Bodnar input lag tester as well as a comparison with a CRT-monitor.
The new test method generates a lot of new data. For now we have chosen to add and show this data in its entirety via our database. As a direct result our pages with the test results of monitors have become a lot longer. We have tried to keep it as clear as possible by showing the data in a different order. The reaction times and input lag will be shown first, followed by the power consumption. After that you will find a small block with the basic results of brightness and contrast, followed by a larger block with the uniformity results. Next up are the viewing angle values and following that are the colour measurements in the default view. If the monitor has an sRGB-mode, an identical block will follow with the results of the measurements using that mode; the same will be done for a possible AdobeRGB-mode.
What do you compare a unique monitor with? Difficult question. In the end we chose 12 different 'regular' superwide monitors, with resolutions of 2560x1080, 3440,1440 and 3840x1600; with the exception of the last category all of them have at least 100Hz refresh rate. In terms of price some of them cost half as much as the Samsung C49HG90, others come closer to the 1287 pounds / 1449 euros - but all of them do not cost as much. So be it.