Originally Posted by HDTVChallenged
Strobing or scanning the backlight at least makes some degree of sense to my feeble brain ... (Yes, I do realize that it was impossible to implement solution before LED backlights.)
Short Reply: Actually, CCFL strobing has long been done. It just wasn't efficient.
Long Reply Below:Re: Sample-and-hold science references, as the proof/basis of this reply
The sample-and-hold science is already well known (e.g. Science & References
) and it has already been shown that mathematically -- sample-and-hold motion blur is equal to the length of the visible frame. Shorter frames is either done by more Hz or via black periods between Hz. And strobe backlights does the job of adding black periods between Hz, so that's the science being convered here. In other words, 60fps@60Hz with 50%:50% dark:bright strobes (excluding other variables, e.g. phosphor decay), have the same amount of motion blur as 120fps@120Hz sample-and-hold (flicker-free). The mathematical relationship is already known.
In addition, sample-and-hold motion blur is demonstrated using a specially-created motion animation: www.testufo.com/#test=eyetracking
(Animation is perfect on a fast GPU-accelerated computer in IE10+ or Chrome, on a CRT, and then view again on common LCD. Newer iPad's also work at 60fps with this animation!)Re: CCFL scanning/strobe backlights
Actually, they did strobing on CCFL backlights, but efficiencies were terrible (e.g. only 20% improvement, 30% improvement) and therefore was not noticeable. It did nothing to close the massive motion resolution chasm between LCD and Plasma. So LCD is forced to rely on interpolation to help things along.
Efficiencies in CCFL strobing was hurt by the following:
-- Long phosphor decay in CCFL on/off cycles. The slow fade-off-and-on of CCFL prevented completely hiding the LCD pixel transitions.
-- If you didn't turn off the whole backlight all at once (e.g. scanning backlight), you get massive backlight diffusion between on segments versus off segments. The backlight leakage prevented completely hiding the LCD pixel transitions.
-- Pixel transitions of older panels were too slow to permit full-strobes instead of sequential-scanning backlights. The slowness meant pixel transitions weren't finished before the backlight strobe. Sequential-scanning backlights enforces the backlight-diffusion limiting factor.
This technological improvement is easily seen in high-speed video of LCD's
-- High speed video of 2007 LCD
-- Refreshes bleed too much into each other. Hitting theoretical efficiencies not possible.
-- High speed video of 2012 LCD
-- Very clear refreshes. Hitting theoretical efficiencies now possible.
This makes possible unbounded motion-blur-eliminating potential (motion blur fully controlled by the backlight, LCD response now a non-factor once squeezed into less than the length of a refresh and hidden by the black period between strobes.) The invention of 3D required LCD panels that could finish refreshing a frame before the next refresh began, which indirectly also made CRT-quality motion resolution possible on LCD.Re: Successfully reaching theoretical efficiency of interpolation-free strobe backlights
These problems have finally been solved in newer LCD panels (fast enough to transition pixels for shutter glasses 3D), as well as combined with newer LED backlights, so of a sudden, technology is finally able to achieve theoretical efficiencies of motion blur reduction by strobe backlight
, as found in LightBoost
(computer monitors) and Motionflow Impulse
50%:50% dark:bright strobing, once per frame = 50% motion blur reduction = 2x more motion resolution
75%:25% dark:bright strobing, once per frame = 75% motion blur reduction = 4x more motion resolution
90%:10% dark:bright strobing, once per frame = 90% motion blur reduction = 10x more motion resolution
These theoretical efficiency gains are finally only now being achieved today.
But yes, because of LED (and fast-enough LCD panels) these theoretical efficiencies in motion resolution improvement are now being achieved on certain displays (e.g. LightBoost) that manage to successfully hide the vast majority of LCD pixel transitions (99%+) in the "off" cycle of backlight (high speed video
). Strobe, then change LCD pixels in dark, then strobe again, then change LCD pixels in dark again, then strobe again (ad infinitum, 60 times per second for Motionflow Impulse, or 120 times per second for LightBoost strobes). The strobe backlight technique can finally be done highly efficiently in newer 2012-era and 2013-era LCD's with backlights bright enough to compensate for the briefness of strobes. In fact, strobe flashes can even be shorter than the length of the LCD transitions, which means my ASUS VG278H (manufacturer rated at 2ms) actually achieves an true MPRT measurement of 1.4ms when tweaked with ToastyX Strobelight (120Hz, LB enabled, LB=10%), because motion resolution is now dictated by the strobe, rather than by the speed of the LCD panel (as long as the pixel transitions are hidden in total darkness). Meaning, true measured pixel response measurements (*and* human perceived motion clarity) can actually be faster than the LCD panel native pixel response (since the native response is unseen during the dark periods).
Further motion resolution improvement potential of LCD is completely unbounded; when it fully becomes controlled by how brief you can strobe a backlight, once per frame. LED's can turn on/off nearly instantly (and white LED phosphor decay is typically 0.1-0.2ms, and you can use RGB LED's instead, too) Once virtually the whole pixel transition (99%+) manages to "fit" inside the off cycle between strobes, it doesn't matter how fast the LCD is. In a 1/60sec refresh (16.7ms), you can achieve 0.5ms response out of an LCD that's 6ms response, if you're using ~16.2ms dark frames and ~0.5ms strobe flashes. You just finish the LCD's 5ms response speed inside the 16.2ms period of darkness (unseen by human eye). Then flash the backlight once the pixels are settled. That's what the human eye sees. A brief flash of a fully sharp, fully refreshed frame. Repeat every frame: You're strobing the backlight on 60fps@60Hz (or 120fps@120Hz) creating CRT-motion clarity (and CRT-like flicker too). Voila -- true native picture response (to human eye / measurements) that is over 10 times faster than the LCD panel's native pixel response! Once you've made the LCD pixel transition speed a nonfactor (it just fits well enough in the dark periods), your motion blur is completely fully controlled by the backlight, and we all know that LED's can be flashed on/off very quickly. Right now, LED wattage is the limiting factor, since you lose a lot of average brightness during strobing. (CRT's can flash briefly at 5000cd/m2 at the phosphor dot, even when just producing an average 80cd/m2 image). Nothing stops a manufacturer from adding extra LED's to flash a brighter backlight for briefer time periods, for future CRT-motion-quality LCD's.Re: Using Interpolation to help provent flicker in a strobe/scanning backlight
For reaching theoretical blur-eliminating efficiency, pure strobes require one strobe per frame, which means if you only have 60fps, you've got 60Hz CRT-style flicker. To go beyond (and reduce flicker) you need interpolation, and that is why some Sony HDTV's let you use Motionflow XR 960, which simulates "960Hz equivalence" by first interpolating to 240fps, and then scanning the backlight at 240Hz in a 75%:25% dark:bright duty cycle (which theoretically adds another 4x motion resolution). But if you eliminate interpolation, and try to simulate "960Hz" motion resolution out of "60Hz", by strobing for 1/960sec only sixty times per second, you're going to get much more noticeable flicker. There are several HDTV's that combine strobing and interpolation
. But that's a big problem for computers/games and interpolation-haters.Re: Disabling interpolation from strobe/scanning backlight
Formerly, most TV's did not allow you to disable interpolation while keeping the scanning backlight. However, high efficiency finally make it worthwhile to provide a menu option to disable interpolation from strobe/scanning backlights (and finally provide it in Game Mode), when needing low-latency output that doesn't modify the image. As people already know, Blur Busters likes low-latency high-efficiency interpolation-free strobe backlights a lot. Only a very few of such efficiency currently exist (e.g. LightBoost strobe backlight, and Sony Motionflow Impulse strobe backlight). These are among the only few that currently manage to hit theoretical motion-blur-reduction efficiencies without interpolation -- required to be computer/game friendly. If we had native 120Hz sources like NHK 8K 120Hz, we wouldn't have to worry about the annoying flicker of 60Hz strobing... (UK has had it bad for motion resolution for a long time: The 100Hz TV's have either needed to interpolate the video *or* otherwise destroy motion resolution. It's been a longtime impossibility: You can't eliminate 50Hz flicker while fully preserving motion resolution *without* interpolation). Until then, dimness and flicker is going to be a common problem in interpolation-free strobe backlights. But they are pure, does not destroy the image pixels, has low lag suitable for CRT-like computer/gaming. Brighter LED's and adjustable strobe duty cycles/capacitor decay (phosphor decay simulator) can help soften this to an extent; and should be a future adjustable/tweakable area of customization if possible. This may be a long term area of development.Re: Conclusion
I hope this reply explains CCFL vs LED strobing -- and better explains why scanning backlights and strobe backlights used to be inefficient -- and how the inefficiencies have been solved/eliminated recently thanks to the combination of LED and faster LCD panels. And also why interpolation may continue to be used for a long time (to reduce strobe backlight flicker). Since it's possible to turn on/off an interpolation-free strobe backlight mode (to gain the pros/cons), it is still a strongly recommended feature of all future LED/LCD HDTV's.Edited by Mark Rejhon - 7/30/13 at 1:19pm