REPORT: Input lag
An investigation of the test process used to date for determining the response time of an LCD monitor, known as input lag
By Thomas Thiemann
Translation by Siobhan Hayes
The switch from bulky cathode ray tube monitors to slim LCDs has led to a certain split amongst users in the computer world. The LCDs saved power and space but were initially considerably inferior to CRTs in terms of display quality, both homogeneity and colour display, and also because of their very limited viewing angles; at the same time, LCDs made themselves unpopular with computer gaming and home cinema fans when they first appeared through various lags. Since lately, LCD monitors and other image reproducing instruments are being used more and more in cars in order to increase visibility at night and warn the driver of obstacles in good time, lags in the screen refresh rate also have an effect in daily life, far from the PC.
The colour display and evenness of the illumination of today’s higher quality models are already better than the best CRTs. However, whilst the switching times of the transistors, often described as response times in the data sheets, are known and indicated in a standardised manner as per ISO 13406-2, and can also be tested with reasonable effort, there is an additional value that has been giving us food for thought for some time: the "input lag".
In this report, the test procedure to date for measuring the input lag has been examined in more detail and improved, and an alternative, more precise process has been developed, with a comparison of the results from both processes. Towards the end of the report, we will then attempt to assess the test process used to date and the results achieved with it.
Input lag is used to describe the difference in time between the signal played to the monitor and the corresponding content displayed on the screen.
This difference is caused by the signal processing that takes place in the monitor. This includes the interpolation when resolutions other than the monitor’s native resolution are displayed, overdrive processes, contrast and other colour adjustments for which, for example, matrix transformations must be carried out with ultimate speed. Fundamentally, all processes where the image information played to the monitor is processed in the monitor can cause a delay in image output and contribute to the overall delay or input lag.
Here, the market relevance cannot be sneered at, since this value is often seen by many monitor buyers as the most important value in deciding whether a monitor is suitable for computer games. Here, a large input lag is generally seen as bad.
This lag adds to the other latencies which arise during image processing and output and thereby causes the information to be delivered with a slight delay in situations where the response time of the user is important, thereby making it harder to respond in time. Transferred to the automotive industry and range of vision optimisations, a slow monitor could reduce the remaining response time of the driver and therefore cut the possible braking distance.
With the methods used to date, input lags of 0 ms to 68 ks have been measured.
At the outset, let us say that at the moment, there is no ISO or other standard for determining the input lag. Thus, no standardised and precisely defined measurement process can be used to create comparable values and be applied industry-wide or privately.
Currently, there is only one widespread process for determining the input lag, which has probably become established because of its low cost. This simple process, which shall be referred to as the "photo method" from this point, is as follows:
An LCD is set up beside a CRT. Both monitors are connected to one graphics card and display a stopwatch in clone mode, i.e. with identical screen content, at a refresh rate of 60 Hz. The stopwatch displays individual milliseconds in the last decimal places. The input lag of the LCD is determined by simply taking photographs of the monitors and measuring the difference in the times displayed.
Examples of the old method for determining the input lag: Left CTX EX1200 vs. Samsung HM2493 with Flash-based stopwatch and right CTX EX1200 vs. BenQ V2400.
Until now, no source has been known to check the accuracy of this process, although the values measured in this way have had significant effects on the purchasing decisions of a wide range of people. With this process, there are a wide range of obstacles which will need to be examined closely in order to allow us to make a statement on whether or not this process provides useful results.