The truth is we're both right.
Originally Posted by Phase700B
I;m not sure your description and analogy comparing strobed back lighting to the scanned line technology of CRTs is accurate. The only "impulses" used in a CRT are those used to hit a particular pixel or spot on the phosphor screen to excite an illuminate it. All CRTs operate in a scanned fashion whereby individual lines of an image are scanned either in an interlaced odd/even fashion or sequentially in a complete progressive scanned frame. This is much different than how a flat panel LCD or plasma works where a whole frame of picture and pixel information is received, dumped into a memory buffer, and then displayed an entire frame at a time on the screen. The "strobed" back light is more analogous to the shutter system another poster mentioned a few posts above.
We're describing scanning backlights from different perspectives, describing separate advantages. From your perspective, you are correct when you're displaying film-based material with a scanning backlight. If you like the "projector look", then you will like a scanning backlight with 24fps movies. No disagreement, that's an advantage for such people who like the projector look. (Note: CRT projectors, driven at 48hz or 72Hz, also help provide a 'projector look' to projected films, too (48Hz is more accurate to a real projector, but flickers more). I used to own an NEC XG135 CRT projector
, and often drove it at 72 Hz with 3:3 pulldown, and also used a PowerStrip tweak to do 48 Hz, too.)
However, I'm also right, for full-framerate material. To me, and other people (in the know), the big benefit of a scanning backlight is LCD motion blur reduction on interactive sources (games and computer) because it's the only possible practical method of motion blur reduction/elimination that (if properly designed) can avoid adding noticeable input lag to video games and computer use. Interpolation is more problematic for games and computers, due to input lag and artifacts.
Also, to understand how scanning backlights can reduce motion blur, it's useful to study these scientific references to understand the 'other' benefit of scanning backlights better (motion blur reduction):
"Dynamic-Scanning Backlighting Makes LCD TV Come Alive.
by Seyno Sluyterman (InformationDisplay.org, October 2005)
"LCD motion-blur analysis, perception, and reduction using synchronized backlight flashing
by Xiao-fen Feng (Sharp Labs. of America Inc., February 2006)
"Frame Rate conversion in the HD Era
by Oliver Erdler (Stuttgart Technology Center, EuTEC, Sony Germany, 2008)Page 4 has very useful motion blur diagrams, comparing sample-and-hold versus impulse-driven displays.
"Perceptually-motivated Real-time Temporal Upsampling of 3D Content for High-refresh-rate Displays
by Piotr Didyk, Elmar Eisemann, Tobias Ritschel, Karol Myszkowski, Hans-Peter Seidel (EUROGRAPHICS 2010 by guest editors T. Akenine-Möller and M. Zwicker)Section "3. Perception of Displays" (and Figure 1) explains how LCD pixel response blur canbe separate from hold-type (eye-tracking) motion blur.
by Stephen Macknik, Barrow Neurological Institute (Scholarpedia)Background information that relates to how flicker becomes a continuous image (applies to CRT and to scanning backlights).
by Michael Kalloniatis and Charles Luu, Webvision (University of Utah)Background information that relates to human vision behavior and how multiple flicker events, over a short interval, blends together.
"Motion portrayal, eye tracking, and emerging display technology
by Charles Poynton (1996) Although this paper is fairly old, it accurately explains eye tracking effects and how it relates to motion blur.
Another way to explain what is Samsung CMR 960 or Sony Motionflow XR 960:
New FAQ entry about Samsung CMR 960 now added to Scanning Backlight FAQ
:Q: What is Samsung CMR 960 or Sony Motionflow XR 960?
The numbers represents a standard motion clarity equivalence to a "X fps @ X Hz" display. These proprietary names/trademarks, attached to these standard numbers, are used by some existing HDTV's with scanning backlights
, and are sometimes viewed as marketing exaggerations by some reviewers. Measurements often show that they do not reflect real-world benchmarks (e.g. contrast ratio claims versus actual measurement).
However, there's a honest actual scientific basis behind these numbers (see Science & References
). These motion equivalence factors are more honest in describing motion blur reduction (under certain conditions) than using "Hz" terminology. Science has shown that motion blur reduction is directly proportional to the length of impulses. Many scientific tests have shown that halving the length of strobe impulses per refresh, halve eye-tracking-based motion blur. Therefore, the shorter the strobe per refresh, and the longer the black period between refresh, the more motion blur reduction occurs. Motion equivalence factors are, in theory, directly comparable to each other on different displays, provided certain assumptions are followed.
[Edit note: This is to equalize the purpose of CRT strobes and LCD strobes (For the proper kind of motion-blur-elimination scanning backlight). This is another interpretation of the exact same formula, that is actually much simpler, provided certain assumptions are followed. See below for a list]
The formula is very simple:
. . . motion equivalence factor = 1 / length of strobe
The honesty of the formula, relative to actual measured science, assumes the following:
How this applies to Samsung/Sony "960" displays:
- One impulse per display point (pixel) per refresh, similar to CRT.
Actual number of backlight strobe impulses can sometimes be more than one per refresh on certain types of scanning backlights (misrepresented factor). Backlight diffusion between adjacent scanning backlight rows, can also lead to multiple impulses for a given pixel reaching the human retinas (unintentional factor). Such factors reduces measurable motion resolution, because multiple impulses are equivalent to repeated frames.
- Full frame-rate material (e.g. 60fps @ 60Hz, or 240fps @ 240Hz)
No repeated frames in the material, because repeated frames leads to increased perceived motion blur caused by eye-tracking. Thus, scanning backlights reduces motion blur during 60fps video games and sports broadcasts (e.g. hockey, football, NASCAR, red bull air races) far more than film-based material (e.g. 24fps non-interpolated). For video games, the graphics must be fast enough to do full frame rate (e.g. 60fps not 30fps).
- Source material is not the limiting factor in motion blur
Video taken with a slow shutter speed, often have built-in motion blur. Overcompressed video also have built-in blur, too. To ensure these are not limiting factors, the camera shutter speed must be faster at the source, than the length of the impulse at the destination display, and the video should not contain visible compression-related motion blur. For video games, artifical GPU motion blur effects should be disabled.
Many displays using a "960" equivalence uses 240Hz refresh, combined with a scanning backlight that's dark 75% of the time. The LED impulse length is 1/960 of a second, with a period of darkness of 3/960 second between impulses (strobe duty cycle of 1/240 second). This results in 1/(1/960) which produces a motion equivalence factor of 960. The purpose of also doing a high interpolated framerate (240fps) is triple fold: It allows more impulses per second without needing a brighter backlight; it reduces scanning backlight flicker (240 Hz flicker instead of 60 Hz flicker); and it reduces stroboscopic effects. Also, other factors above, affect actual perceived motion blur reduction, such as backlight diffusion between adjacent scanning backlight sections.How this applies to CRT:
It is already well known that 60fps @ 60Hz on a CRT, have much clearer-looking motion than even 240fps @ 240Hz on a LCD. This formula explains why CRT has far less motion blur than LCD -- a CRT display has approximately a 1 millisecond phosphor decay
. Such a display has motion fluidity that looks the same, to human eyes, as "CMR 1000" or "Motionflow XR 1000", or a 1000fps@1000Hz display! No wonder CRT motion is so sharp, even at only 60Hz!
Some additional notes:
- Long-persistence CRT's obviously have more motion blur than short-persistence CRT's. This is compliant with the formula above.
- Have you operated an old film pre-war-era projector that used one shorter light impulse per frame? (Those had to use shorter light impulses, due to slower film frame movement). Those have less motion blur (on sufficiently fast-enough-shutter-recorded film), provided the projector isn't vibrating/shaking the film much! This also happens to be compliant with the formula above, too!
- Have you seen an image transmitted through a Nipkow disc? Ala 1920's television experiments? (Modern hobbyist nipkow discs use LED's instead of neon lamps, google "LED nipkow"). They are really low spatial resolution, and very dim, but have excellent temporal resolution -- better than CRT due to zero decay after pixel (spatial) illumination. No motion blur in moving objects; and scrolling text on nipkow discs are not perceptibly blurrier than stationary text. These are compliant with the formula above too!
- Scrolling LED marquees that use line-driven LED pixels (non-capacitored) have excellent temporal resolution. Strobe lengths are so short on some models (and espeically older red-color marquees), when you track your eyes on these signs, the LED's look like they are moving along with the text. Virtually zero eye-tracking-based motion blur. These are compliant with the formula above too! Adding a capacitor to the LED's lengthens the strobes, and you see more eye-tracking based motion blur on the individual LED's in scrolling text on marquees that do not flicker (or do not use short strobe lengths). This is ALSO compliant with the formula above too!
In conclusion, the scanning backlight design needs to match the behaviour of CRT strobes (as described above), to gain the same benefits of motion blur elimination, as CRT's. Otherwise, if the assumption is violated, there is little/no motion blur elimination for a scanning backlight. So you leave behind only other advantages (e.g. "the repeated frame projector look") that scanning backlights can do. Newer scanning backlights are doing a better and better job at eliminating motion blur, thanks to technological improvements. Also, on newer LCD displays where pixel persistence/ghosting is far less than a single LCD refresh, it is now possible to use a strobed/scanning backlight to bypass the LCD pixel persistence as a limiting barrier to motion blur, simply by turning off the backlight for the entire pixel persistence period. (see diagrammatic page explaining how, here
It is quite important to note that real-world scanning backlights do not always meet the above list of assumptions in the FAQ entry, and certain scanning backlight technologies (older scanning backlight technologies that only do imperceptible/meager (e.g. 10%) real-world motion elimination). The actual measured blur reduction is far away from the actual mathematically calculated motion blur reduction. So both you and I rightfully dismiss many of these early attempts as gimmick
, but the science & theory of a scanning backlight design, for major motion blur reduction, is scientifically sound
(and many clearly do see the benefit in CRT's in material meeting the critera listed above). Scanning backlights can be engineered for massive motion blur reduction/elimination. I believe we also need increased awareness about the benefits to games/computers that scanning backlights can provide (without the input lag disadvantage of motion interpolation). Recent motion resolution tests have shown that we are gradually approaching that era of less gimmickly- designed scanning backlights, that have a strobe duty cycle that much more closely resembles CRT impulses than yesterday's scanning backlight technologies. Improved motion tests need to be done. The media education about scanning backlights is extremely poor, and many early models are rightfully gimmick (see! I agree with you here)
, but it's certainly quite nonsense to dismiss all of them as gimmick. The ones that far more accurately emulate an impulse-driven display, are the ones with real motion blur reduction.Edited by Mark Rejhon - 10/27/12 at 2:22am