Doug Kerr
Well-known member
Two of the monitor tests conducted with test images in the Lagom suite are a bit obscure (although there is quite a nice bit of background given on the site for the second one):
• The Clock and Phase test
• The Inversion (pixel-walk) test
I myself had to do some research to fully learn their significance, and I thought I would pass on my findings. Both pertain only to LCD-type displays (the thrust of the Lagom test suite).
The Clock and Phase test
In a CRT-type monitor, for each color channel we basically send (over the analog "VGA" interface) a video signal to paint all the pixels in "raster" fashion. There are no discrete "pixel sites" on the screen, so the screen just displays the different luminance values (for the different pixels) as they arrive, wherever the "scan" is at that instant.
An LCD display has fixed, discrete pixel sites. With a digital interface (DVI), the pixel values are sent serially, along with a clock so that the "receiver" in the monitor can keep up with when the voltage on the interface means one pixel or the next.
But likely our LCD monitor can also be operated from a traditional VGA interface. Now, the "receiver" in the monitor must "strobe" the waveform at the center of the instant when each pixel value is sent to capture that pixel value and put it into the memory map of the image. Since no clock accompanies these signals, the receiver must deduce from the waveform on each interface lead the exact frequency and phase of the needed clock (just as we do when receiving a serial data signal via a modem). The adaptive systems for doing this are very sophisticated.
Nevertheless, some monitors have user controls to "tweak" their clock recovery systems for most reliable performance with the exact clock rate of the "video card" that feeds the monitor on a VGA interface.
The "Clock and Phase" test pattern provided in the Lagom suite is intended to be "challenging" for the clock recovery system. Thus, with it in place, we can adjust the clock frequency and phase controls on the monitor (if it has such) to tweak it for most reliable operation.
The Inversion (pixel-walk) test
In an LCD, a voltage is applied across each little cell of the panel which controls the transparency of the liquid crystal content of the cell.
But the continuous appearance of a voltage of the same polarity will quickly "polarize" the cell, leading to misbehavior.
Fortunately, a voltage of either polarity can be used to control the transparency of the cell, both polarities having (almost) identical effect.
Thus, we could, on every alternate refresh cycle, use alternate polarities for the required voltages for each cell (the "inversion").
But in fact, the response of the cells to the two polarities may not be identical, and so the exact magnitudes of the voltages have to be set to compensate for that, else there will be a flicker of the display at half the refresh rate. And evidently it is quite challenging to get the proper voltages for each transparency level.
To ease this situation, a scheme is used in which some pixels will be in their "positive" voltage state at any given refresh cycle and others at their "negative" state, with the situation reversed for the next refresh cycle. (Or maybe this will even be done by "tricolor dot".) Thus, over any small area of the display, the average "flicker" (as perceived by the viewer) will be zero. There are a number of different schemes used for this purpose.
But, for any given scheme, we can contrive pixel patterns that will "thwart" this strategy (they only "turn on" pixels or dots in one of the two polarity phase groups), and thus make flicker appear.
In the Lagom inversion (pixel-walk) test, there are contrived patterns that will thwart six different widely-used schemes, leading to visible flicker.
Some of the schemes organize the two groups of pixels (or dots) in pairs of rows, or pairs of columns. The exact pattern needed to thwart the flicker avoidance must take this into account. But we often don't know how the pattern in the monitor starts, nor can we be certain how the test pattern will be "laid on the screen" (owing to such things as menu area on our "viewer").
Thus, for the patterns to thwart such strategies, the Lagom suite provides two patterns (e.g., 4a and 4b) with different row (or column) phases.
Thus if our monitor has a flicker-avoidance strategy that can be thwarted by "pattern 4", for the current "lay-down" of the test pattern, the flicker will show up on either panel 4a or 4b.
What is the point of having this in the Lagom suite? Beats me. In most cases, there is nothing we can adjust to deal with less-than-theoretical performance in this regard. (Some monitors have internal adjustments fir the voltages.) Perhaps what we will see is flicker for several patterns, which tells us that the scheme is not working as well as it theoretically can.
On thing we can tell from which of the patterns causes visible flicker is which of the flicker-avoidance schemes our monitor uses. Except that the "key" to this is not given by Lagom. (In some other test suites that provide this test, the relationship is stated.)
Then term "pixel walk" perhaps refers to the fact that for certain kinds of image patterns (e.g. fine cross-hatch patterns), there can be an interaction between the image pattern and the effect of the flicker-avoidance scheme that produces a sort of "crawl" of the image pattern.
Further reading
My grasp of these two mysteries largely came from this rather nice article on LCD displays and their testing, by William Andrew Steer, PhD, of techmind.com (the originator of the inversion test principle used in the Lagom suite):
http://www.techmind.org/lcd/
It includes test patterns for the two matters discussed in this note.
Further good information is here:
www.intersil.com/data/an/an1208.pdf
a paper also cited by Nienhuys.
Best regards,
Doug
• The Clock and Phase test
• The Inversion (pixel-walk) test
I myself had to do some research to fully learn their significance, and I thought I would pass on my findings. Both pertain only to LCD-type displays (the thrust of the Lagom test suite).
The Clock and Phase test
In a CRT-type monitor, for each color channel we basically send (over the analog "VGA" interface) a video signal to paint all the pixels in "raster" fashion. There are no discrete "pixel sites" on the screen, so the screen just displays the different luminance values (for the different pixels) as they arrive, wherever the "scan" is at that instant.
An LCD display has fixed, discrete pixel sites. With a digital interface (DVI), the pixel values are sent serially, along with a clock so that the "receiver" in the monitor can keep up with when the voltage on the interface means one pixel or the next.
But likely our LCD monitor can also be operated from a traditional VGA interface. Now, the "receiver" in the monitor must "strobe" the waveform at the center of the instant when each pixel value is sent to capture that pixel value and put it into the memory map of the image. Since no clock accompanies these signals, the receiver must deduce from the waveform on each interface lead the exact frequency and phase of the needed clock (just as we do when receiving a serial data signal via a modem). The adaptive systems for doing this are very sophisticated.
Nevertheless, some monitors have user controls to "tweak" their clock recovery systems for most reliable performance with the exact clock rate of the "video card" that feeds the monitor on a VGA interface.
Remember when television receivers had manual horizontal and vertical frequency (horizontal and vertical hold) controls?
The "Clock and Phase" test pattern provided in the Lagom suite is intended to be "challenging" for the clock recovery system. Thus, with it in place, we can adjust the clock frequency and phase controls on the monitor (if it has such) to tweak it for most reliable operation.
The Inversion (pixel-walk) test
In an LCD, a voltage is applied across each little cell of the panel which controls the transparency of the liquid crystal content of the cell.
But the continuous appearance of a voltage of the same polarity will quickly "polarize" the cell, leading to misbehavior.
Fortunately, a voltage of either polarity can be used to control the transparency of the cell, both polarities having (almost) identical effect.
Thus, we could, on every alternate refresh cycle, use alternate polarities for the required voltages for each cell (the "inversion").
But in fact, the response of the cells to the two polarities may not be identical, and so the exact magnitudes of the voltages have to be set to compensate for that, else there will be a flicker of the display at half the refresh rate. And evidently it is quite challenging to get the proper voltages for each transparency level.
To ease this situation, a scheme is used in which some pixels will be in their "positive" voltage state at any given refresh cycle and others at their "negative" state, with the situation reversed for the next refresh cycle. (Or maybe this will even be done by "tricolor dot".) Thus, over any small area of the display, the average "flicker" (as perceived by the viewer) will be zero. There are a number of different schemes used for this purpose.
But, for any given scheme, we can contrive pixel patterns that will "thwart" this strategy (they only "turn on" pixels or dots in one of the two polarity phase groups), and thus make flicker appear.
In the Lagom inversion (pixel-walk) test, there are contrived patterns that will thwart six different widely-used schemes, leading to visible flicker.
Some of the schemes organize the two groups of pixels (or dots) in pairs of rows, or pairs of columns. The exact pattern needed to thwart the flicker avoidance must take this into account. But we often don't know how the pattern in the monitor starts, nor can we be certain how the test pattern will be "laid on the screen" (owing to such things as menu area on our "viewer").
Thus, for the patterns to thwart such strategies, the Lagom suite provides two patterns (e.g., 4a and 4b) with different row (or column) phases.
Thus if our monitor has a flicker-avoidance strategy that can be thwarted by "pattern 4", for the current "lay-down" of the test pattern, the flicker will show up on either panel 4a or 4b.
What is the point of having this in the Lagom suite? Beats me. In most cases, there is nothing we can adjust to deal with less-than-theoretical performance in this regard. (Some monitors have internal adjustments fir the voltages.) Perhaps what we will see is flicker for several patterns, which tells us that the scheme is not working as well as it theoretically can.
On thing we can tell from which of the patterns causes visible flicker is which of the flicker-avoidance schemes our monitor uses. Except that the "key" to this is not given by Lagom. (In some other test suites that provide this test, the relationship is stated.)
Then term "pixel walk" perhaps refers to the fact that for certain kinds of image patterns (e.g. fine cross-hatch patterns), there can be an interaction between the image pattern and the effect of the flicker-avoidance scheme that produces a sort of "crawl" of the image pattern.
Further reading
My grasp of these two mysteries largely came from this rather nice article on LCD displays and their testing, by William Andrew Steer, PhD, of techmind.com (the originator of the inversion test principle used in the Lagom suite):
http://www.techmind.org/lcd/
It includes test patterns for the two matters discussed in this note.
Further good information is here:
www.intersil.com/data/an/an1208.pdf
a paper also cited by Nienhuys.
Best regards,
Doug
Last edited: