The Kelvin scale has always been relatively simple to understand in the context of display calibration. Choose a higher value and your display white point will be bluer; choose a lower value and it will be yellower, etc. But this is where the simplicity stops and the confusion begins. What do we do if the white point appears to be magenta or greenish? And why doesn’t the display calibration Kelvin value correlate with the value of our lighting? In other words, why does calibrating to ~5700 Kelvin match 4100 Kelvin lights? Why can’t we calibrate to 4100K to match 4100K lights? So hard for new users…

The Kelvin scale rendered on a chromaticity diagram
The Kelvin scale rendered on a chromaticity diagram

For years, I’ve encouraged people to choose a Kelvin value that allows for the whites on-screen to match paper white under their critical print lighting, using 5700K as a starting point. Picky users like myself often end up switching to more precise controls that allow for X and Y color axis control points. This way demanding users wanting a a better paper white match on-screen can make small adjustments adding or subtracting any hue in the spectrum.


There are lot of monitors and software used for display calibration, and I can’t go into all of them here. NEC and others have had great tools for tweaking the white point after measurement in real time. Most programs require that you make separate profiles with different white point settings. 

 

i1Profiler's display calibration with xy White Point setting selected
i1Profiler’s display calibration with xy White Point setting selected

 

For those using XRite’s i1Profiler to calibrate their monitors, let me suggest x=0.329 y=0.339 as a new starting point. Once you’ve calibrated and compared the white’s onscreen to paper white, you can repeat the process tweaking the white point x y cross hairs using the intuitive chart.  Increasing the y value makes the white greener – decreasing y makes them more magenta, etc. A single click in either direction is significant so make small changes of less than 10 between calibration attempts. Save these x and y values in the name of your display profile so you can refer to them later. 

Well calibrated displays should have matching paper whites, both in terms of color and brightness. Once you use more precise white point controls that allow for X and Y color axis control points, I think you’d never want to think about display calibration white point settings in terms of Kelvin ever again. IMO, it’s time we move on from that.