So let’s look at some of the data we’ve collected at SpectraCal.
Not surprisingly, different types of displays show distinctly different luminance properties over time. Let’s start with a standard LCD. Below is a graph of sequential luminance readings taken over a 70 minute span. These are the Y values in foot-lamberts over the time in minutes.
From our everyday experience, this is probably just what you would expect the curve of the luminance change to look like. It is clearly nonlinear but is very intuitively understood. It starts off dim then quickly gets brighter but eventually levels off to an almost constant luminance.
[...]
A better way to put it is that we want the percent difference over a given duration to be zero, i.e. no change in luminance. In the first 10 minutes there is a 235 percent luminance change. From 10 to 20 minutes it drops down to a smaller but still significant change of 6.19 percent. At 100 IRE this difference probably cannot be perceived by the human eye. But it is important to remember that our vision's perception of luminance is logarithmic and is far more sensitive to low level light changes than high ones. So a change of 6 percent in total light output can drastically affect our black levels. We need this change to be much lower. Below is a table of percent changes over 10 minute spans.
[...]
At SpectraCal, we will never calibrate a meter until our reference display stabilizes to a change in luminance of less than 0.1 percent per minute. That means that over a 10 minute span, we need to see less than 1 percent of change. On this LCD, that doesn't occur until somewhere after the 50 minute mark. Of course, if you're calibrating a client's display, you may not always have time to wait 50-plus minutes and you'll have to use your own discretion in determining what change in luminance is acceptable. However, if time is not an issue, I would recommend waiting at least 60 minutes.
Now let's look at some other display types. Below are two warm up graphs: one for an LED backlit display, and one for a CRT display
I was extremely surprised to see that as these displays warm up, they actually stabilize to a dimmer state. This means that if we adjusted our black levels too soon, there may be a "crushing" effect after the display stabilizes. These displays stabilize to a luminance change of less than 0.1 percent per minute after 40 minutes for the CRT and after
70 minutes for the LED.
