Originally Posted by lipcrkr
Well, in my everlasting quest to determine the cause of motion artifacts in the 71 series 120hz panel, one cause i mentioned ions ago was that the 8ms spec is too high for a 120 hz panel. This is an interview with Brian Berkley who is with Samsung (more specific info below) with HDTV magazine. The new Samsung panels in the 6 & 7 series are spec'd at 4ms. Could this be the holy grail? Could the lower spec'd circuits work more compatably with AMP? Why did Samsung cut the 8ms in half? Did they talk to Brian Berkley?
Liquid Crystal Pixel Response Time
The short answer to the 4ms part of the question is:
The timing is the response time in ms that a single liquid crystal pixel of a display device takes to twist and turn itself to change from white to white passing thru black first. Manufacturers also declare the spec from gray to gray, from black to black, to express the same thing, and could be shorten with "G-G or g-g".[/i]
In other words, the timing for a liquid crystal pixel to restore itself and be ready to display the next image without showing any residue of the previous image displayed by that pixel (otherwise artifacts could occur).
In theory, the faster a liquid crystal pixel can restore itself, the better potential the display device would have to display more frames per second (faster frame refresh rates), for whatever purpose.
Some manufacturers express the spec as the number of ms the liquid crystal pixel takes to reach only up the halfway point (ie. measuring from white to black, before converting to white again). If my memory serves me correctly, several years ago Sony specified some SXRDs LCoS (Liquid Crystal on Silicon) sets with about 2.5ms for that half cycle when introduced, which in theory would mean 5ms for the full cycle. .
In my exchanges about the subject with Brian Berkeley, Vice President, LCD Business Technology Development Group, Samsung Electronics, the concept was further clarified:"G-G response time can be reduced (a good thing) by application of response time compensation (RTC) circuits, which compare gray levels of the previous and current frames to determine a boost value.
This boost, or overshoot, causes the LC molecules to switch to the proper position more quickly. Samsung's version of RTC is called dynamic capacitance compensation, or DCC. For an LCD-TV to be free of motion blur, it is a requirement that the g-g response time be small enough."
The frame refresh rate is a specification of the incoming signal but is also of a display device, which could have by design the capability to show 60 frames per second of incoming video at faster speeds if necessary, such as 120Hz or 180Hz. What kind of content is displayed on those extra frames is another subject, more on that later.
Although that spec is a timing of the video engine, not the liquid crystal pixel, the overall performance of the set could be affected by the interaction of both. For example, the response time of each pixel could be insufficient when is much slower (taking longer to restore itself) than the speed required by the frame rate of the display. In simple terms, having a slow 20ms pixel response time on a 120Hz display (that would require at least 8ms of pixel speed) would render an inferior image.
In other words, the set intends to display video frames at a speed that the pixels response time cannot catch up with, affecting the pixels readiness for the next frame of video, and show blurriness, lag, etc. which is a common problem on poorly designed LCD panels.
Typical LCD panels show each frame in a "sample-and-hold" manner, by which all the pixels on each frame are kept lighted for the whole duration of the cycle for that frame rate (16ms for 60fps), and at the very end of that cycle the crystals twist and turn in a rush to be ready and adapt to the next frame, which also brings full light with it. That is done even when the pixel could have restored itself faster at its own independent response time of 2, 4, 6, 8 ms depending on the design.
In other words, the blurriness is not always caused because the liquid crystal pixel response time is slow, but because there might not be a break of light in between frames.Note that here you can capitalize from the comment I made earlier about frame refresh time ('Note that I am not saying "the timing that takes a full frame to be displayed", which is a different concept.')
As follows: The frame could be displayed really fast within the 16 ms period between frames, and still have the time to DISAPPEAR out of sight to show black or something else to interrupt light, rather than sit there fully lighted and on hold until is time for the next frame.
An important factor is that, to perceive continuous motion, the human eye expects the image frames to have breaks in between and interrupt the constant supply of light, so it might not be misinterpreted as increased blurriness.
The sample-and-hold approach for each frame does not provide for that break, and when the next frame needs to be shown the twist-and-turn of the liquid crystal pixels is done so quickly that is not seen as a light interruption.
While discussing this subject with Brian Berkeley, he confirmed the following:"Hold type driving is used in LCD-TVs and in other matrix-addressed displays. Compare this type of driving with CRT displays, which are impulsively driven. In an impulsively driven display, the image is present only for a short period of time. By contrast, in a hold-type driven display like an LCD, the image is held throughout the entire frame period until next frame is written.
Hold-type driving is good for eliminating flicker and for getting maximum light output, but there is a drawback to hold-type driving for moving pictures. If an image is moving at a rate of, say, a few pixels per frame, then there is effectively a positional error in that the image position is correct for only a fraction of the frame.
To measure motion picture response time of an LCD panel, there is a metric called "MPRT", which literally stands for motion picture response time. MPRT is useful for comparing different LCD-TVs, but it is not so useful for comparing LCDs to other technologies, such as plasma displays, projection displays, or CRTs. As in g-g time, it is also good to have a lower MPRT score.
Important note: Having very small gray-to-gray response time does not solve this problem. Even if the g-g response time is 0, MPRT can still be too large to have blur-free images. Again, this is due to hold-type driving mentioned above. A low g-g LC response time is necessary, but not sufficient, for achieving a low MPRT score."
Faster Frame Rates
If a display's frame rate is faster than the incoming signal, it has the potential to smooth out the motion and improve the presentation of images. For example, 24fps video that originated from film content will show flicker if displayed at that speed, some displays can accept that 24fps frame rate and display the frame cadence at exact multiples of it to eliminate the flicker and to give smoothness to fast action images, without doing 2:3 pull-down to convert the film frame rate to the more typical 60fps of video.
Some displays multiply the frame rate x2, x3, etc. to display the 24fps content at 48Hz (front projectors), 72Hz (Pioneer Elite plasmas), 96Hz (front projectors), and even at 120 Hz (some projectors, LCDs), creating or adding new frames that were not present in the incoming signal.
Interpolate New frames
To impart smoothness to the fast movement of the objects shown by the image, some sets insert new frames employing interpolation techniques that calculate its pixels anticipating the future direction of the content motion by looking at the next frame/s in advance, rather than just repeating the same exact frame.
Such technique claims to improve the presentation of fast images, but I add, provided that the video processing is of sufficient quality to avoid adding artifacts that might render a cadence of images that is poorer than just repeating the same frame.
Such repetition would be more faithful to film content, as the local theater projector do with film, showing the 24 film frames per second at 48 frames speed by opening the projector's shooter twice for each celluloid frame that passes in front of the projector lens.
The 120 Hz Competition
Having a spec of 120Hz on a LCD display does not necessarily mean that the set is doing pixel-by-pixel motion adaptive interpolation of added calculated frames as mentioned above, it depends on the design, and it does not necessarily mean that it accepts 24fps film sources frame rate to display them at 5 times that speed, whereby 1 incoming frame is repeated 4 times before displaying the next frame.
Some might use the faster 120Hz display speed to give smoothness to the 60Hz video source presentation, and convert 24Hz sources to 60Hz with 2:3 pull-down, and then double the frame rate to 120Hz of the display by repeating twice the frame (rather than the 4 repeated frames technique above). Because of the conversions, there could be a lot of video artifacts compared to straight frame repetition.
Some LCDs are inserting black or darker (same image at lower light) frames within the 120fps display cadence of an incoming 60fps image. Some can do both (black/darker frames, or image interpolated frames) at the option of the viewer and selectable from the video menu of the set.
Some can show black intervals between the actual 60fps frames without having a 120Hz speed rate design, such as just interrupting the light source in between frames, rather than adding black frames that need to be displayed at a faster frame rate.
The insertion of black (or darker) frames or light interruptions between actual frames is viewed by many manufacturers as a good method for the human vision to better perceive motion from LCD with less motion blur, having a clean separation of video frames with no (or reduced) light source in between, rather than the sample-and-hold approach mentioned before.120Hz is becoming the new buzzword for LCD panels to give consumers the idea of "my panel is faster and is therefore better than the competition", but many fall short in describing how that speed is used. The image cadence could be faster but at the price of interpolation artifacts from a sub-standard video processor, to keep the panel price down, which also attracts consumers.
Manufacturers like JVC, Toshiba, Sharp, Philips, Samsung, Hitachi, and LG have implemented proprietary technologies such as Clear Motion Drive II, ClearFrame, http://www.tacp.toshiba.com/televisi...y.asp?fid=2088
, FineMotion Advanced, and Pixel Plus (from the first four manufacturers above respectively), to address the sample-and-hold issues, improve motion smoothing, and reduce the perceived motion blur. They employ methods of adding black frames, interpolated frames, darker frames, backlight flashing, at 60, 120 or 180 Hz frame rates.
Brian Berkeley, further adds to this concept:"g-g and MPRT scores are being confused with one another. There is nothing particularly great about a g-g response time of 8ms; 4ms g-g is more typical these days. On the other hand, an MPRT score of 8ms is about state-of-the-art for mass production LCD-TVs these days.
So how to overcome the problem of hold-type driving? By using 120Hz driving with frame interpolation, instead of 60Hz driving, this effectively cuts the hold time in half. If the g-g response time is low enough, the MPRT score will also be cut in half.
Another technique is to use black insertion, which causes the LCD to mimic a CRT's impulsively driven response. But the problem of applying such kind of impulsive driving technique to LCDs is that (just like CRTs) flicker can occur, and there is light loss.With this as background, yes, a 120Hz set must have g-g response times below 8ms."
Mine is a Super Fast Pixel
Why pursuing a shorter 2-4-6 ms pixel response time when it would not solve the blurriness problem by itself? Most LCD panels are designed with frame rates of 60fps, and now with 120 fps to address the problem, and those theoretically would need respectively 16 or 8 ms pixel response time.
Maybe those pixel-fast panels were preparing for soon to be implemented technology of faster frame rates, which would put to use such fast speed at the pixel level. Maybe they should have both been implemented together to claim that the blurriness problem in LCD was properly addressed as a package.
In the middle of those question marks there is an issue of market competition and "be first in technology" by a manufacturer, and how that played a role in convincing consumers that 2-4-6 ms response time was actually needed in a world of mostly 60Hz panels, and even now with 120Hz panels.
Brian Berkeley responded to me with one final comment about that issue:"Yes, 120Hz driving needs faster g-g response time. The calculation of required g-g isn't trivial; it depends on several factors. For example, for a 120Hz panel, it isn't quite as simple as saying the g-g response must be within 1/120Hz or 8.33ms.A g-g response below that number is required. Also, g-g is just one number, but how it gets measured is important: Does g-g response time mean average g-g over all possible transitions, does it mean the max value of all possible transitions, is it measured at 10%-90% or some other transition %, and so on.
The question is then: how much lower is low enough (lower than 8ms for 120 Hz panels) to make sure the g-g as measured is sufficiently low (fast) for the requirement of the panel frame rate? Considering that the g-g measurement standard is not typically disclosed on a g-g spec other than saying "x ms", and even if it would be, I doubt most people would know what to do with it.
By design the g-g should be low enough, but are you the type of consumer that believes that manufacturers do not cut corners believing you won't notice? Or one that rather go a bit lower than 8ms g-g, just to be on the safe side, maybe down to 4ms, if that is a safe zone for your pocket.Would you buy a Ferrari with 55 Mph rated tires, or rather with 200 Mph rated tires?
Even when law limits the 55 Mph speed, you might feel better because one day such high performance on that piece of the car "might be needed".