Andy Rosen from Bitlogic and Dan Brunner from Microsoft gave a compelling demonstration in the Display session at the SMPTE conference where they showed some interesting artifacts that they say are attributable to the overdriving techniques used in nearly all LCD displays to improve the speed of response. The effect has been long suspected, but the two now say they have test patterns that can demonstrate the measureable artifacts.
The motivation for the team’s work was to determine if an LCD monitor was suitable for high frame rate video editing. But along the way, they discovered some perplexing and disturbing artifacts generated from test patterns and even real life content.
Below is a black and white image of a tunnel with concentric rings. Beside it is a close up of the image played back on an LCD monitor. Notice the color artifacts introduced in this monochrome image! Rosen and Brunner think this is an artifact of the LCD overdrive circuits as the effect is not visible on OLED displays or DLP monitors. LCDs have a hard time with slightly off-vertical monochrome patterns.
But is gets worse. In the pattern below, a clock hand rotates around 4 four squares that are supposed to be at a balanced D65 white. Notice that the middle of the Chroma Shift Wipe pattern3 contains four color patches. These are used as a guide to judge the severity of the white shift of the monitor under test. On some LCD monitors, you may notice that the degree of white balance shift is greatest when the sweep hand is at noon. Other monitors exhibit the problem for the full rotation of the sweep hand. They have even seen one LCD monitor that recovers by the time the hand reaches two o’clock.
Rosen and Brunner noted that these shifts in the color patches is measureable, so a real problem exists here, not just a visual perception issue.
In a live demo, they showed a moving monochrome zone plate on the 3-chip DLP projector and on an LCD monitor. There was no color shift on the projector, but noticeable color bands on the LCD monitor.
You can download the patterns yourself to investigate further at http://alturl.com/gs3ou. I did and when running the moving zone plate pattern above on my monitor, I did not see any color fringing. Same thing for the rotating clock. Does that mean the monitor does not use overdriving?
