or Connect
AVS › AVS Forum › Display Devices › Display Calibration › Standardizing Motion Resolution: "Milliseconds of motion resolution" (MPRT) better than "Lines of motion resolution"
New Posts  All Forums:Forum Nav:

Standardizing Motion Resolution: "Milliseconds of motion resolution" (MPRT) better than "Lines of... - Page 3

post #61 of 69
Thread Starter 
Quote:
Originally Posted by Elix View Post

Err... I am in trouble converting it into ms of motion blur. Does it mean those optomas outperform all other projectors in terms of motion resolution (except CRTs)? What number of "lines of motion resolution" equivalent would that be?
Yes, it does.

Typical 60 Hz LCD or DLP with black frames disabled/interpolation disabled -- about 16.7ms of persistence (even if GtG is only 1ms or 2ms)
Typical 120Hz LCD or DLP with black frame disabled/interpolation disabled -- about 8.3 of persistence
Adding black frames to 120Hz -- about 4.1ms of persistence (for 50%:50% dark:bright duty cycle).
I'm ignoring interpolation; as that's not good for gaming.

This translates to: During 1000 pixels/second motion (1 millisecond per 1 pixel), you get a minimum 16.7 pixels of enforced display motion blur (during eye tracking) during framerate=Hz motion during video games (full frame rates) for 60fps@60Hz on a sample-and-hold 60Hz. For 120Hz, that's 8.3 pixels. For 120Hz+traditional BFI, that's 4.1 pixels of motion blurring respectively. Note, that GtG curves (finite time that a pixel takes to transition from one color to another) will generally 'add' to persistence; so these numbers are ideal case scenario, assuming continuous light output. This is observed when doing motion tests such as www.testufo.com or PixPerAn, and observing the changes in motion blur trail length (this is observed with LightBoost monitors, see Photos of motion blur: 60Hz vs 120Hz vs LightBoost.

Black frames can vary away from 50%:50% duty cycle. For example, black frame insertion could be 3:1 ratio (dark 75% of the time, bright 25% of the time). Such black frame insertion ratios would reduce motion blur by 75% instead of 50%. Likewise, a 90%:10% dark:bright duty cycle would reduce tracking motion blur by 90%. (these numbers are only exact if black frame insertion is perfectly square-wave; generally it doesn't always necessarily reach these efficiencies. However, DLP's do highly efficient black frame insertion since they turn off pixels extremely fast.).

-- Most LCD/LCoS projectors don't use black frame insertion, so they generally more motion blur than DLP's.
-- CRT's generally have approximately ~1ms of persistence from the phosphor illuminate-and-decay cycle (less for shorter persistence, more for longer-persistence). Although not a squarewave persistence, the brightest part of the illumination cycle is often at sub-millisecond levels, so that part determines the human-perceived motion clarity. That's more than 15x sharper motion clarity than a 60Hz sample-and-hold LCD, so the bar is set extremely high for displays that attempt to match CRT motion clarity.
post #62 of 69
Great paper from 2010 (open access) that has some useful figures:

(notice the difference in scales in the first figure - one is in microseconds, the other in milliseconds.



post #63 of 69
Thread Starter 
Yes -- that's right, some CRT's illuminate-and-decay a few hundred microseconds (back to levels too dim to create significant stimulus), while other CRT's take a lot longer (~2ms). The Sony GDM-W900 CRT is one of the longer-persistence phosphors, taking ~2ms.

Also the LCD curves visually resemble an older 5ms LCD. The curves of a 1ms gaming LCD are more cliff-shaped than those, but still otherwise similar.
post #64 of 69
Mark, you have so much passion about this as I see almost each of your posts is long yet informative. How do you do this?) Keep it up. smile.gif

So, basically, Optoma GTs (720-760) are not as good as Lightboost monitors in terms of motion resolution? About 2-4 times more blur? And here I was aiming at ordering one for 3D gaming.
Maybe next year we'll see not only G-Sync monitors but G-Sync projector from Acer. smile.gif Looking forward for that.
post #65 of 69
Ok, in the images below, the green box with the arrow to its right is the one that is moving. Slightly above it are two adjacent objects separated by a black dashed line (this dashed line is not visible during the trial, but included here for illustration). The object on the right is the "base". By itself, it represents what the moving square would look like with zero motion blur. To its left is a static representation of a simulated blur trail. Using the indicated keys, the observer modulates the shape of this trail.

In each image, the top graph represents the function that describes the shape of this simulated trail. The observer uses the keys to modulate certain parameters of the function in real time. The function is then applied to the trail and changes its appearance in real time. When a match is made, the user presses a key to exit, and the output is the bottom graph. This represents the pixel decay function as inferred from the shape of the trail.

The top image uses a hybrid of two functions: an exponential function to describe the rapid decay of phosphor luminance, and a linear portion to describe its "tail". In reality the linear portion is actually more of a power law, but for ease of programming, I used a linear function. The user can adjust both the base of the exponential portion, and the slope of the linear portion.

The bottom image uses a cumulative normal distribution. Here, the two parameters are mean, and sigma. The mean is the X axis value at which Y = 50%. Sigma denotes the steepness of the curve. A sigma of 0 would essentially be a square wave. In the example shown here, the mean is 0.9, and sigma is 0.03 (relative to a function that goes from 0 to 1 along both axis).

Note that in both images, the displayed function is an actual representation of the displayed simulated trail. These trails, however, do not match what I actually visually perceived under these pixel speed conditions. I adjusted the parameters of the function to values that would easily illustrate what is going on.

Also note that during the trial, the graphs, arrows, etc. are not visible.

The hybrid function is probably more suited to describing a CRT display, while the cumulative norm may be more suitable for strobed backlight displays. I'm planning to add a third and fourth variable to the cumulative norm function. I've noticed that at least some strobed displays have a blur trail that is faint but at a fixed lower contrast relative to the object that's creating the trail. This creates a faint ghosting of the moving object. By allowing the user to change both the baseline of the function (so that instead of going from 0 to 100 luminance it can go from, say, 30 to 100), and the width of entire function (representing the width of the trail), I think the "step type" function can be more adequately characterized.

I also found that the halation effect made this task difficult on a CRT (this is when bright pixels create a glow around them), as it masks what is really going on in both the actual blur trail and the simulated blur trail. I think this sort of approach is better suited to stroboscopic displays.

I am fairly confident that if the appropriate function is found for a given display type, this approach might have success, especially with higher MPRTs (higher MPRTs means more visually extended blur trails, which means easier to match).

Hybrid:




Cumulative Norm:


Edited by spacediver - 10/30/13 at 1:06am
post #66 of 69
Quote:
Originally Posted by Elix View Post

Mark, you have so much passion about this as I see almost each of your posts is long yet informative. How do you do this?) Keep it up. smile.gif

So, basically, Optoma GTs (720-760) are not as good as Lightboost monitors in terms of motion resolution? About 2-4 times more blur? And here I was aiming at ordering one for 3D gaming.
Maybe next year we'll see not only G-Sync monitors but G-Sync projector from Acer. smile.gif Looking forward for that.


You won't see lightboost projectors anytime soon because if you cut into ON time with BFI to gain OFF time, you also cut into the grayscale perfomance and that's going to hurt (going below 8 bits ).

You'd need significantly better light engine to make up for the loss of time spent dithering. Also , cycled colors won't allow stellar motion resolution either.

(Lack of subpixels on the other hand is a plus).
post #67 of 69
Quote:
Originally Posted by Mark Rejhon View Post

-- Sony's Motionflow Impulse on certain Sony TV's (dramatic horserun ahead of all other LCD's). People have reported being surprised by the clarity of 60fps gaming on these.

Sony did it.., the mistake of the year... their 'smart' engineers decided to enable their interpollation algorithm (after the latest firmware updates) at 'Impulse' mode, and there is no way back.... you can't disable it now....OMG. frown.gif

Impulse mode was the only MotionFlow option where the Sony 4K had the best motion resolution from any other LED in the market, with NONE Interpolation.... sadly now.. Sony's Impulse mode features bad interpollation, large motion blur and this mode now is just a dimmed mode that noone will ever use... frown.gif
post #68 of 69
Perhaps outdated and out of place for this discussion, but here's how the video experts who wrote the Final Technical Report (1995) for the ATSC HD system summarized their testing of HD motion resolution: Table 2.3 shows target and measured static/dynamic resolutions for luma and chroma at 5 rpm for motion. Back in '01, I posted a simplified table of the measured resolutions, which now needs a "control A" for legibility with my Windows browser after the recent AVS archive processing; the old tech-report URL shown was discontinued, too. -- John
Edited by John Mason - 1/29/14 at 8:11am
post #69 of 69
No matter how much you post about this, no matter how long and informative posts are, it is still IMPOSSIBLE to characterize motion blur ON ANY video display, and especially on LCD displays, because the blur is different under different conditions.

Some combinations of colors blur more than others. Some combinations of luminance blur more than others. You could look at some content and see/measure very little blur, while other combinations would make you think the display has the worst blur on the planet.

So there is NO single number, not even six single numbers, that would characterize blur with any meaning no matter HOW it is measured.

And do you really care if one LCD display has 80% blur (not that it can be characterized by a single number) while another LCD display has 65% blur? Both suck and isn't that really the bottom line? EVERYBODY who is into video displays KNOWS that plasma has FAR less blur than LCD. And a lot of people may know that DLP has extremely low blur. So WHY BOTHER?????? You have "bad" blur from LCD, fairly good blur performance from plasma, and excellent blur performance from DLP. If LCD displays or projectors (frankly, I don't see much blur in LCoS digital cinema projectors and am not sure why since LCoS home theater projectors have PLENTY of blur, but then, LCoS cinema projectors do 3D without blur but home theater LCoS projectors still aren't completely ghost-free with 3D).

To characterize blur you'd need to measure 100%, 75%, 50%, 25% and 10% versions against at a minimum the 6 primary and complimentary colors and those primary and complimentary colors would also have to be measured at 100%, 75%, 50%, 25%, and 10% plus you'd have to do black/gray/white evaluations against black/gray/white and against the primary and complimentary colors also at 100, 75, 50, and 10. And you'd have to do the colors moving against stationary black/gray/white fields. The math makes my head hurt. But this would take something like 960 measurements to characterize a display's blur but that's so much data it's meaningless. All you need to know about this topic is what technology is not good, what tech is moderately good, and what tech is the best. That's pretty easy to figure out in 10 or 15 minutes of looking at the motion patterns that come with Display Mate with Motion. Does anybody really care if something sucks "80" or something sucks "65"? They both suck and if neither one is good enough for you, get some other display tech that doesn't have so much blur.

Hard to understand why this subject keeps going on and on when it doesn't really mean much. We can SEE whether a display has blur well enough to tell if it ia bad good or excellent. And you really don't need anything more than the motion patterns in DisplayMate with Motion or maybe on the Spears & Munsil version 2 disc to tell if a display is bad, good, or excellent.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Display Calibration
AVS › AVS Forum › Display Devices › Display Calibration › Standardizing Motion Resolution: "Milliseconds of motion resolution" (MPRT) better than "Lines of motion resolution"