Originally Posted by xrox
It is that simple. There is only one flicker on a plasma per refresh created by the subfield weighting sequence. You are not understanding correctly. Individual subfields do not produce a percievable flicker. It is the weighted sequence that does.
That's what I am talking about.
We are just talking about different timescales. Your eyes may not, but my eyes can perceive the individual subfields as separate flickers during certain motion tests.
Some flickers are brighter than others. It's often a sequence where many flickers are imperceptible, but a few flickers in certain colors (e.g. the last few subfield pulses of a refresh, the brightest ones) might be visible enough to be a dominant, and in certain cases one can see the temporal displacement of the subfield flickers. The last few flickers of a pulsed sequence can be detected by the human eye as separate distinct flickers, since subfield flickers often reveal themselves as a distinct artifact of banding (approximately PWM-style) in the motion blur of things during fast motion. This also varies on color, e.g. mixed colors, dark shades (can appear as multiple-flicker-per-refresh effect if you're sensitive enough) versus solid whites (resembles a clean one-flicker-per-refresh effect). For black/white solids (e.g. moving solid white object on solid black background), several plasmas can make all flickers dim and only make the last flicker very bright (it shows in high speed camera), since in that case, the plasma doesn't need to add colors via the temporal dithering effect. But for certain colors, there's obvious multiple flickers, and I can see (with human eye) motion artifacts caused by the multiple visible subfield flickers, e.g. blending into artifacts in fast-moving murky colors, along sharp versus fuzzy edges. The flickers can do a kind of a temporal dithering effect (very noticeable in dark shades), since the flickers of some pixels are offset from the flickers of other pixels, in order to create intermediate shades. Plasma subfields are very complex, even neither you, nor I, nor Chronoptometrist, can fully understand (especially when you also throw in extra things like motion interpolation into the subfield refreshes -- which the Panasonic VT50 definitely does). These noticeable flickers average out to a single flicker to the human eye. Isn't this exactly the same thing we're talking about? We're just interpreting plasma differently, but we're singing to the same choir with similar intentions. It's often just a matter of different terminology / semantics, which is creating misunderstandings here.
Also, noticed your subsequent post. It seems we are talking about exactly the same thing. Pulses that ramp up in brightness, that essentially merge into a single flicker. This is consistent with what I see in high speed camera pointed at a plasma.
An observation, motion blur math is not simple with plasmas, due to all this subfield complexity that varies a lot from plasma to plasma, plus the red/green phosphor decay. (Unlike for pure clean strobes, it is hard to predict in advance exact motion blur on a plasma in advance without doing actual motion measurements). Just voice me the strobe length of a LCD strobe backlight (not scanning backlight, due to diffusion interference), as long as it successfully creates a black frame much longer than the LCD's rated pixel transition speed, the LCD backlight strobe length accurately predicts amount of motion blur. But the motion blur math is not as simple with plasma. By being told the strobe length, one can blindly predict the amount of motion blur on a full-strobe-backlight LCD (all-at-once flashes eliminate backlight diffusion from interfering with motion blur math) as LED backlights have no perceptible phosphor decay (<0.2ms), and find that motion tests show darn near exactly the predicted amount of motion blur. But I can't do that with plasma. So many variables (the ramp-up of subfield refreshes, the combined "temporal dithering effect" of the subfield refreshes causing different MPRT measurements in different colors, the phosphor decay, etc.) The motion blur math on plasmas is not simple enough to predict human-perceived plasma motion blur in advance. Not a knock on plasma, but I'm just saying it's "not that simple"
Yes, from the posts I read, it seems we already agree that there's one dominant flicker per refresh, as the multiple adjacent flickers blend out to an average single flicker, since the flickers are often dim until the last few flickers of the subfield in consecutive flickers that continually ramp-up in brightness, so it's as if the flickers are clustered together, essentially visually blending into one single perceived flicker per refresh as seen by the human eye.
Again, I am very sensitive to plasma flicker and when I walk up to the plasma, I can on occasion, see the temporally-displacement of different subfield flickers individually during certain motion tests with certain color boundaries (colors that aren't easily done with only one bright subfield pulse with all pixels in the same subfield, rest of subfields invisible/dark). Likewise, with DLP when I look closely, I can still see rainbows in high-speed 6X (360Hz) color wheels if I intentionally look for these. With any tech (CRT / DLP / strobe backlight LCD / plasma), flickers do not give me a headache, just that I've been able to identify motion artifacts caused by high frequency flicker. Just as many people can be trained to detect 3:2 judder or trained to detect rainbows, I've trained my eyes to be able to detect >1000Hz flicker indirectly via the Phantom Array Effect (Scientific References
-- scientific tests show up to 10Khz is indirectly detectable using this technique). I also can see temporal dithering in 3-chip DLP during an intentional very fast eye movement that is perpendicular to a high-contrast edge -- this is harder to detect than color-wheel artifacts. For seeing temporal artifacts on a display, one good test is a moving line or edge that runs very fast across the screen (e.g. TestUFO: Blur Trail, or other motion tests that presents moving edges/lines, and can also be configured to intermediate non-full-dark/non-full-bright colors) running at different colors at various different pixel rates, to distinguish the multiple subfields. It separates the weak plasmas from the strong plasmas. Also, under a high speed camera (I have a budget 1000fps camera which I've pointed at a few displays), the multiple flickers are often temporally different colors; which indicates a lot of plasma take advantage of the temporal time & spatial displacement to do both temporal dithering and spatial dithering simultaneously. I appreciate the plasma engineer's complex technical decisions that are made to create a plasma that creates nicely solid colors. I'm again, saying single flicker per refresh is "not that simple".
But yes, we agree, the temporal displacements of the multiple pulses essentially merge into one nice looking, colorful flicker per refresh.
That said, if you point a high speed camera, each pulse appear as consecutively-brightening dithered-look flickers (as seen in individual frames of 1000fps camera
) at the millisecond timescale, that blends into to a single colorful flicker at the sixteen-millisecond timescale. We are just simply thinking of different timescales. At the 16-millisecond timescale, yes, yes, it's a single nice-looking flicker. But at the sub-millisecond timescale, it's multiple noisy-looking off-color flickers seen in individual frames of a high speed camera, and revealable to the human eye via certain configured computer-driven motion tests. Even when watching certain video material, I can even see artifacts caused by the multiple temporally-displaced subfields, even moreso when I'm at computer-monitor-view distance from plasmas (1:1 screen width viewing).
Perhaps I am using wrong terminology. I'm worthy of blame for that. I've just chosen the terminology that most accurately describes what I've seen with the camera and eyes. I'm viewing plasma without knowing the plasma engineering terminology, but with a technological appreciation of the temporal complexity of a plasma display, and observing with my eyes (that is sensitive to artifacts caused by various temporal effects). So in that sense, we are both correct -- just from different perspectives/different timescales/artifact sensitivity/viewpoints. It's fairly simple looking at the 16ms timescale (the "weighted" term you said -- the average flicker is the whole series of subfields), but very complex at the sub-millisecond timescale.
Just as people are often trainable to detect 3:2 judder, or trainable to detect rainbow artifacts, even many eyes can actually be trained to see temporal dithering artifacts in both DLP and plasmas, in certain motion tests such as TestUFO: Moving Line, 2pix thick, 960pps, 25% Cyan, 75% Yellow
, or TestUFO: Moving Line, 8pix thick, 1920pps, 25% Gray, Light Magenta
. (View these in a supported browser
such as Chrome on a recent GPU-accelerated computer or laptop, run browser in full screen mode -- such as a fairly recent ATI/nVidia accelerated laptop, make sure everything in chrome://gpu is green). Try it on older plasmas and newer plasmas. One or the other test, or both, will often show a distinctive noisy blur trail. Better plasmas will do better at keeping this motion clean, but can begin to fail during more complex situations (e.g. softer murky gradients rather than hard edges). The noisy stuff you see is your human eyes successfully seeing the temporal effects of the subfield refreshes, and the banding stuff your eyes see, is the temporal separation of the individual flickers of the subfield refreshes. Play with non-full-dark and non-full-bright colors using color selectors, till you find a combination that creates a noisy or banded blur trail; you're witnessing motion artifacts caused by sub-millisecond-scale temporal behavior of the pixels in the display. View these motion tests at 1:1 view distance or slightly closer, approximately the relative viewing angle of a computer monitor or a projector home theater. They have very noisy-looking motion blur trails on some plasmas, if you go right up to the plasma and track your eyes on this line; some of them create a motion blur that show distinct banding artifacts that is caused by by the temporal displacement of the individual subfield flickers. Obviously, if you view a plasma display from regular sofa view distance (2:1), the artifacts are far less visible, and often not visible anymore.
I think all of us, can agree on, that plasma display engineers are very smart. Getting finicky gas-filled cells to display a beautiful home theater picture with minimum possible motion artifacts.Edited by Mark Rejhon - 7/25/13 at 7:38pm