The problem of a lag in displaying frames on some monitor models has been widely discussed on Web forums recently and seems to resemble the response time issue. However, it is quite a different effect. With ordinary fuzziness, the frame that the monitor receives is displayed at the same moment, but it takes some time for the frame to get drawn up fully. With the input lag effect, some time, a multiple of the monitor’s refresh rate, passes between the arrival of the frame from the graphics card and the moment is begins to be displayed. In other words, the monitor is equipped with a frame buffer – ordinary memory – that stores one or several frames. Each frame arriving from the graphics card is first written into the buffer and only then is displayed on the screen.
It’s simple to measure this lag. You should connect two monitors (a CRT and an LCD one, or two different LCD monitors) to two outputs of the same graphics card in Clone mode and then launch a timer that would count up milliseconds. Then make a series of photographs of the monitors’ screens. Then, if one monitor has an input lag, the showings of the timers in the photographs would differ by the value of the lag: one monitor is displaying the current timer showing, and the other is showing a value that has been true a few frames earlier. To get reliable results, you should make a score of photographs and discard those of them that obviously fell at the moments the frames changed. The diagram below shows the results of such measurements for a Samsung SyncMaster 215TW monitor (in comparison with an LCD monitor that has no input lag). The X-axis shows the difference in the showings of the timers on the screens of the monitors. The Y-axis shows the number of frames with the given difference.
A total of 20 photographs were made and 4 of them obviously fell at the moments of changing the frames (they showed two values on the timers, one from the older frame and one from the new frame). Two frames gave a difference of 63 milliseconds, three frames – 33 milliseconds, and 11 frames – 47 milliseconds. Thus, the correct lag value for the 215TW is 47 milliseconds or about 3 frames.
As a kind of digression, I want to say that I’m rather skeptical about some posts at forums whose authors claim an abnormally low or high input lag on their particular sample of a monitor. They just don’t accumulate enough statistics and make only one frame. As you have seen above, separate snapshots may catch a lag either higher or lower than the real value. The higher the camera exposure value, the higher the probability of such a mistake. To obtain real data, you need to make a couple of dozens of snapshots and then choose the most frequent lag value.
Well, this is all irrelevant for our shopping needs. We won’t go photographing timers on monitors we want to buy, will we? From a practical standpoint, the question is if this lag should be bothered about at all. Let’s take the above-mentioned SyncMaster 215TW with its input lag of 47 milliseconds. I don’t know of monitors with a higher input lag, so this example makes sense.