Many people believe that motion blur in LCD displays is due to the fact that liquid crystal pixels take a relatively long time to change colour. In fact, response times have very little to do with blur levels. With the right backlight strobing techniques, a 120Hz LCD display with a response time of 8ms is entirely indistinguishable from one with a response time of 1ms. The barrier separating traditional LCD and AMOLED displays from the motion clarity of CRTs and PMOLED displays is persistence, not response time.

Left: An object moving across a low persistence display in slow motion. Right: An object moving across a full persistence display, also in slow motion.
Left: An object moving across a low persistence display in slow motion. Right: An object moving across a full persistence display, also in slow motion.

Persistence is a measure of how long each image sent to a display stays on that display. In short, lower persistence leads to better motion clarity. When an object in an image moves across the image, our brain expects it to do so smoothly. With a full persistence display (where each image stays on the display until the next image arrives), this is not the case. What is shown is that the object “jumps” across the screen every update, and stays still for the duration of each frame. For instance, when we are halfway between frame 1 and frame 2, our brain expects to find the object halfway between its position at frame 1 and its position at frame 2. However, since the display has not updated yet, the object is still at its location in frame 1. This mismatch between predicted position and actual position causes motion blur.

CRTs have practically zero persistence. As the electron beam scans down the rows of the display, each pixel is lit for only a moment, before the beam moves on to stimulate another phosphorescent pixel, and it no longer produces light. This gives them the advantage of practically perfect motion clarity. When we are halfway between frames, searching with our eye for a moving object, we instead find black. Instead of finding incorrect information about the location of the object, as we would with a full persistence display, we find no information at all until the next frame. This alleviates motion blur.

LCD displays typically use full persistence. The backlight shines through the liquid crystals, which together with polarised filters, colour each pixel. The backlight remains on constantly and the liquid crystals remain (practically) in the same configuration for the duration of each frame. The juddering quality of this full persistence display technique is what causes the majority of motion blur in LCD displays, not the time taken for each pixel to switch from one state to another with each new frame.

NVIDIA 3DVision uses active 3D glasses. An LCD shutter in each lens blocks light from the display to each eye alternately, in sync with the display’s refreshes. During even frames, the left shutter is closed, so the frame is seen only by the right eye. During odd frames, the right shutter is closed, so the frame is seen only by the left eye. The problem with this system is that the LCD shutters do not open and close instantly. There is a period between frames where both shutters are partially open, and the user experiences crosstalk, where the left eye sees part of the image meant for the right eye and vice versa.

To combat crosstalk, the second version of 3DVision included a technology called LightBoost. LightBoost is a strobing backlight system for 3DVision certified LCD displays. The backlight stays off most of the time, turning on briefly, once every frame. The strobing is synchronised with the shutter glasses, so that the backlight is turned off while the shutters are opening and closing. This way, users see pitch black between frames, eliminating crosstalk. Of course, this strobing happens at a frequency far beyond the flicker fusion point, so we do not notice it at all. LightBoost has the consequence of “hiding” the image while an LCD display is switching, making response times irrelevant to motion blur. It also gives LCD displays low persistence, eliminating persistence based blur too. Enthusiasts used software to trick LightBoost capable monitors into thinking they were running in 3D mode, enabling LightBoost all the time for perfect motion clarity. Since then, monitor manufacturers have been implementing their own low persistence technologies into LCD monitors.

OLED displays fall into one of two categories, passive matrix (PMOLED) and active matrix (AMOLED). PMOLEDs emit light only when they are updated, much like pixels in a CRT which emit light only when the electron beam is incident on them, providing zero persistence. AMOLED displays use a transistor backplane to provide constant power to the OLEDs, making AMOLED a full persistence technology. AMOLED achieves better energy efficiency and display lifetime than PMOLED. To achieve the same perceived brightness as a pixel in an AMOLED display, a PMOLED pixel needs to flash on at a much higher power, to compensate for the fact that it emits no light for most of the frame. Driving OLEDs with higher power, like many electronic components, reduces their efficiency and lifespan.

Left: An undistorted car sprite. Right: The sprite may appear sheared when moving across a display with sequential refresh.
Left: An undistorted car sprite. Right: The sprite may appear sheared when moving across a display with sequential refresh.

One would think that PMOLED is the obvious choice for VR displays, where motion clarity and contrast are king. However, this is not the case. The Oculus Rift and HTC Vive both use AMOLED displays, yet still achieve low persistence. How? They strobe backplane power, much like NVIDIA’s LightBoost technology strobes backlight power in LCDs. The reason for this is that in a PMOLED display, each frame is “smeared” over time. Rather than arriving in one flash, every update is rolled out sequentially down each row. This results in fast moving objects exhibiting horizontal shearing. By using a strobed backplane, VR displays achieve low persistence and global refresh, where the entire frame is shown at once, meaning no shearing can occur.