Display lag is a phenomenon associated with some types of LCD displays, and nearly all types of HDTVs, that refers to latency, or lag measured by the difference between the time a signal is input into a display and the time it is shown by the display. This lag time has been measured as high as 68ms, or the equivalent of 3-4 frames on a 60 Hz display. Display lag is not to be confused with pixel response time.
Analog vs digital technology
For older analog cathode ray tube technology, display lag is extremely low due to the nature of the technology which does not have the ability to store image data before display. The picture signal is minimally processed internally, simply for demodulation from a radio frequency carrier wave (for televisions), and then splitting into separate signals for the red, green, and blue electron guns, and for timing of the vertical and horizontal sync. Image adjustments typically involved reshaping the signal waveform but without storage, so the image is written to the screen as fast as it is received, with only nanoseconds of delay for the signal to traverse the wiring inside the device from input to the screen.
For modern digital signals, significant computer processing power and memory storage is needed to prepare an input signal for display. For over-the-air or cable-TV, the same analog demodulation techniques are used, but after that the signal is converted to digital data which must be decompressed using the MPEG codec, and rendered into an image bitmap stored in a frame buffer. This frame buffer is then procedurally written to the display device. In its simplest form this processing may take several microseconds to occur.
Causes of display lag
While the pixel response time of the display is usually listed in the monitor's specifications, no manufacturers advertise the display lag of their displays, likely because the trend has been to increase display lag as manufacturers find more ways to process input at the display level before it is shown. Possible culprits are the processing overhead of HDCP, DRM, and also DSP techniques employed to reduce the effects of ghosting - and the cause may vary depending on the model of display. Investigations have been performed by several technology related websites; some of which are listed at the bottom of this article.
LCD, plasma, and DLP displays, unlike CRTs, have a native resolution. That is, they have a fixed grid of pixels on the screen that show the image sharpest when running at the native resolution (so nothing has to be scaled full-size which blurs the image). In order to display non-native resolutions, such displays must use video scalers, which are built into most modern monitors. As an example, a display that has a native resolution of 1600x1200 being provided a signal of 640x480 must scale width and height by 2.5x to display the image provided by the computer on the native pixels. In order to do this while producing as few artifacts as possible, advanced signal processing is required, which can be a source of introduced latency. Interlaced video signals such as 480i and 1080i require a deinterlacing step that adds lag. Anecdotally, display lag is significantly less when displays operate in native resolutions for a given LCD screen and in a progressive scanning mode. External devices have also been shown to reduce overall latency by providing faster image-space resizing algorithms than those present in the LCD screen.
Many LCDs also use a technology called "overdrive" which buffers several frames ahead and processes the image to reduce blurring and streaks left by ghosting. The effect is that everything is displayed on the screen several frames after it was transmitted by the video source.
Testing for display lag
Showing the existence of input lag requires a test display (the display being measured), a control display (usually a CRT) that would ideally have no display lag, a computer capable of mirroring output to two displays, stopwatch software, and a high-speed camera pointed at the two displays running the stopwatch program. The lag time is measured by taking a photograph of the displays running the stopwatch software, then subtracting the two times on the displays in the photograph. This method only measures the difference in display lag between two displays and cannot determine the absolute display lag of a single display. CRTs are preferable to use as a control display because their display lag is typically negligible. Also, video mirroring does not guarantee that the same image will be sent to each display at the same point in time.
In the past it was seen as common knowledge that the results of this test were exact as they seemed to be easily reproducible, even when the displays were plugged into different ports and different cards, which suggested that the effect is attributable to the display and not the computer system. An in depth analysis that has been released on the German website Prad.de revealed that these assumptions have been wrong. Averaging measurements as described above lead to comparable results because they include the same amount of systematic errors. As seen on different monitor reviews the so determined values for the display lag for the very same monitor model differ by margins up to 16 ms or even more.
To minimize the effects of asynchronous display outputs (the points of time an image is transferred to each monitor is different or the actual used frequency for each monitor is different) a highly specialized software called SMTT or a very complex and expensive test environment has to be used.
Several approaches to measure display lag have been restarted in slightly changed ways but still reintroduced old problems, that have already been solved by the former mentioned SMTT. One such method involves connecting a laptop to an HDTV through a composite connection and run a timecode that shows on the laptop's screen and the HDTV simultaneously and recording both screens with a separate video recorder. When the video of both screens is paused, the difference in time shown on both displays have been interpreted as an estimation for the display lag. Nevertheless this is almost identical to the use of casual stopwatches on two monitors using a "clone view" monitor setup as it does not care about the missing synchronisation between the composite video signal and the display of the laptop's screen or the display lag of that screen or the detail that the vertical screen refresh of the two monitors is still asynchronous and not linked to each other. Even if v-sync is activated in the driver of the graphics card the video signals of the analog and the digital output will not be synchronized. Therefore it is impossible to use a single stop watch for display lag measurements, nervertheless if it is created by a timecode or a simple stopwatch application, as it will always cause an error of up to 16 ms or even more.
Effects of display lag on users
Display lag contributes to the overall latency in the interface chain of the user's inputs (mouse, keyboard, etc.) to the graphics card to the monitor. Depending on the monitor, display lag times between 10ms and 68ms have been measured. However, the effects of the delay on the user depend on the user's own sensitivity to it.
Display lag is most noticeable in games (especially older video game consoles), with different games affecting the perception of delay. For instance, in World of Warcraft's PvE, a slight input delay isn't as critical compared to PvP, or to games favoring quick reflexes like Counter-Strike. Rhythm based games such as Guitar Hero also require exact timing; display lag will create a noticeable offset between the music and the on-screen prompts. Notably, many games of this type include an option that attempts to calibrate for display lag. Arguably, fighting games such as Street Fighter and Tekken are the most affected, since they may require move inputs within extremely tight windows that sometimes only last 1-3 frames on screen.
If the game's controller produces additional feedback (rumble, the Wii Remote's speaker, etc.), then the display lag will cause this feedback to not accurately match up with the visuals on-screen, possibly causing extra disorientation (e.g. feeling the controller rumble a split second before a crash into a wall).
TV viewers can be affected as well. If a home theater receiver with external speakers is used then the display lag causes the audio to be heard earlier than the picture is seen. "Early" audio is more jarring than "late" audio. Many home theater receivers have a manual audio delay adjustment which can be set to compensate for display latency.
Many televisions, scalers and other consumer display devices now offer what is often called a "game mode," in which the extensive preprocessing responsible for additional lag is specifically sacrificed to decrease, but not eliminate, latency. While typically intended for videogame consoles, this feature is also useful for other interactive applications. Similar options have long been available on home audio hardware and modems for the same reason.
Display lag versus response time
LCD screens with a high response time value often do not give satisfactory experience when viewing fast moving images (They often leave streaks or blur; called ghosting). But an LCD screen with both high response time and significant display lag is unsuitable for playing fast paced computer games or performing fast high accuracy operations on the screen due to the mouse cursor lagging behind. Manufacturers only state the response time of their displays and do not inform customers of the display lag value.
Input Lag example for console gaming
The process that occurs from when the user presses a button to when the screen reacts is outlined below (steps which have negligible response time contributions have been omitted). Each step in the process adds response time (commonly known as "input lag"), which varies from minor to noticeable.
1: Controller sends signal to console For wired controllers, this lag is negligible. For wireless controllers, opinions vary as to the effect of this lag. It is likely that opinions vary due to each user's sensitivity to lag, model of wireless controller and the other equipment in the signal chain (i.e. the rest of their gaming setup).
2: Network lag (online gaming only) Since the console must know the current location of other players, there is sometimes a delay as this information travels over the network. This occurs in games where the input signals are "held" for several frames (to allow time for the data to arrive at every player's console) before being used to render the next frame. At 25 FPS, holding 4 frames adds 40ms to the overall input lag.
3: Console processes information and sends frame output to television A console will send out a new frame once it has finished processing it. This is measured with the frame rate. Using Gran Turismo 5 as an example, the maximum theoretical framerate is 60 FPS (frames per second), which means the minimum theoretical input lag for the overall system is 17ms (note: the maximum real world FPS in 3D mode is 40-50 FPS). In situations where processor load is high (e.g. many cars are on-screen on a wet track), this can drop to 30 FPS (16 FPS for 3D mode) which is equivalent to 32ms.
4: Television processes frame (image correction, upscaling, etc.) and pixel changes colour This is the "input lag" of the television. Image processing (such as upscaling, 100 Hz, motion smoothing, edge smoothing) takes time and therefore adds some degree of input lag. It is generally considered that input lag of a television below 30ms is not noticeable, discussions on gaming forums tend to agree with this value. Once the frame has been processed, the final step is the pixel response time for the pixel to display the correct colour for the new frame.
Typical overall response times Overall response times (from controller input to display response) have been conducted in these tests: http://www.eurogamer.net/articles/digitalfoundry-lag-factor-article?page=2 It appears that overall input lag times of approximately 200ms are distracting to the gamer. It also appears that (excluding television input lag) 133ms is an average response time and the most sensitive games (first person shooters and Guitar Hero) achieve response times of 67ms (again, excluding television input lag).