So what refresh rate do I need? [Analysis] [very good one!]

Advanced display talk, display hackers, advanced game programmers, scientists, display researchers, display manufacturers, vision researchers & Advanced Display Articles on Blur Busters. The masters on Blur Busters.
User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 17 Feb 2014, 16:22

I think the issue that some forum readers take is the specific, hard, exact number (60) being suggested when it really varies all over the place depending on variables (direct view, peripheral vision, duty cycle, softness of flicker, ambient light, human dependant factor, big display, small display, bright display, dim display, TV viewing distance, desktop viewing distance, etc) that may vary from the said scientific environments (tests/studies) that are done.

But altogether, for the purposes of this thread, we can assume "60" as an elastic number representing "your specific individual direct flicker detectability threshold for a specific viewing situation". 48Hz-50Hz bothered huge numbers of people except in totally dark movie theaters, 60Hz was tolerated by many in television-room use cases, 75Hz-85Hz became the CRT sweet spot for computer-desktop use cases, small segment of population still notice flicker directly even at >100Hz in their viewing environment, etc.

For good virtual reality headsets, ambient light is blocked, and a significant portion of your field of vision is covered, so once your eyes acclimates after putting on the headset, the "above flicker detection threshold" would be approximately 75Hz-80Hz for a lot of people, and this is ballpark targeted by Oculus for their low-persistence prototype. So, the number 75Hz makes a hell of a lot sense (though higher is always better to reduce the stroboscopic effects, as I described.)
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

ScepticMatt
Posts: 36
Joined: 16 Feb 2014, 14:42

Concerning CFF and peripheral vision

Post by ScepticMatt » 18 Feb 2014, 09:14

Looking closer, the first posts only concerns CFF around the fovea ("centrally fixed test stimulus"). After comments about peripheral vision flicker perception, I looked up additional research:

Concerning CFF and peripheral vision:

If I understand correctly, the latest model for CFF (ignoring saturation and scotopic vison) is the following:

f(E,L,d) = (0.24E + 10.5)(Log Li + 1.39 Log d - 0.0426E + 1.09) (Hz)
f = CFF
E = eccentricity in degrees
Li = retinal illuminance in Troland (Td = 10,000 cd/m2),
d = stimulus diameter in degree
with Log Li = Log L + Log p this changes to

f(E,L,d, p) = (0.24E + 10.5)(Log L+log p + 1.39 Log d - 0.0426E + 1.09) (Hz)
f = CFF
E = eccentricity in degrees
L = eye luminance in Troland,
d = stimulus diameter in degree
p = pupil area in mm2
So, let's assume a worst case scenario (correct me if wrong):
Dark adapted pupil at 1cd/m2: 13mm2, log p = 1.1
215cd/m2 screen = 3.45 log Td according to paper
110 degree VR HMD full white, ultra low persistence flicker

centered eye: CFF = 89 Hz
35 degree right/left: CFF = 127 Hz

Source (page 9): http://www.researchgate.net/publication ... 1bf958.pdf
via: http://www.journalofvision.org/content/11/5/13.full.pdf (discusses latest research on peripheral vision)
bonus: graph from paper:
Image

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 18 Feb 2014, 11:01

ScepticMatt wrote:centered eye: CFF = 89 Hz
35 degree right/left: CFF = 127 Hz
That's some good vision research of the extreme cases, pretty close to what I'm observing in real world in situations that roughly resemble these extreme cases. Although not "one size fits all" humans (so many variations in vision and vision defects -- after all, 8% of humans are color blind, according to some statistics -- and some forms of color blindness often reportedly affects flicker sensitivity -- and there are other human specific issues too).

There are apparently human specimens that deviate from this measured model. I know there are several people who can't tolerate staring directly at 120Hz LightBoost flicker (1% of population? Hard to say). It would be rather interesting to run a larger test (e.g. 10,000 humans) to identify the 100 humans with unusual flicker sensitivity, and then do some more tests on these, and attempt to determine what it takes to achieve a display that satisfies these humans. Maybe some of these humans are unusually sensitive to the indirect stroboscopic effect (e.g. minor eye movement & sudden headaches occurs instantly & their brain highlights the stroboscopic effect unusually strongly), or maybe some of these humans actually have much higher flicker detection threshold than the average individual, or maybe some kind of resonance/phenomenon is occuring in their eyes/human brains that aren't seen as flicker but felt as pain -- such as 360Hz PWM creating 360Hz vibrations/voltage inside retinas that bothers brain -- or some other interaction). I'd love to see more scientific studies done on this "1%" sensitive people -- the abberations of humans. When selling 1,000,000 displays, that's still 10,000 million complainers, so it would be statistically significant to a big-time manufacturer (e.g. people who can't tolerate CRT/plasma but love LCD).

Conversely, it would be lovely if more scientific studies were done on the "1%". People who get headaches triggered from motion blur (e.g. people who love CRT but can't tolerate standard LCD). Since Blur Busters started (the world's first website to directly cover strobe backlights for gaming), I've been contacted by more than one dozen people who thanked me profusely for eliminating their headaches. Also, VR makes this a bigger issue -- Oculus have confirmed -- that headaches can occur with motion blur when motion blur is forced upon you because it's unnatural external motion blur. So low persistence VR helps reduce the nausea/headache issue a lot, and some scientific studies are probably starting to be done about this as we speak?? (I'd hope!) Because, apparently it actually also occurs with some people using LightBoost. The existence of this site, has revealed the existence of a small segment of population who get more problems (nausea/strain/headaches) from the motion blur, than from the flicker, meaning there are now increasingly more and more situations (in the modern display era) where motion blur is the bigger evil compared to flicker.

Yesterday, we had slow motionspeed, low frame rates, fuzzy graphics, standard definition, small monitors. But today.... As motionspeeds get faster, graphics get sharper, framerates get higher, more pixels are on a display, display cover bigger span of view -- all of them makes it easier to see/detect motion blur. I immediately notice that VR combines the worst case scenario (fast motionspeed caused by head turning, high framerates of modern GPUs, sharp 3D graphics, high-res textures, high definition displays, wide field of view). Graphics is also much sharper than video. All these variables hugely dramatically raise a human's sensitivity to motion blur forced upon you by a display. No wonder low persistence solves a hell lot of VR headaches problems (I say: duh! I knew this, despite the lack of scientific studies on this). However, people sitting close to large 1080p strobe-backlight monitors who play a heavy amount of FPS, sometimes notice the strain-reducing benefits of low persistence. There is a crossover point where sometimes low persistence is preferred over flicker-free/strobe-free. So it affects display users too, albiet to a lesser extent. (The ones that use extreme fast motion and are more acclimated to it than the average gamer) This is why my LightBoost FAQ has both entries "Q: Why does LightBoost have MORE eyestrain?" and "Q: Why does LightBoost have LESS eyestrain?"

But there seems to be always be that <"1%" of population who are always uncomfortable any form of motion blur or any form of flicker/stroboscopic (even indirectly) -- the people who get immediate headaches with 432Hz PWM (unconsciously or otherwise). I wish we could combine low persistence AND flicker/strobefree (but as me and Michael Abrash and others point out, that requires ultrahigh frame rates), so we could satisfy five-sigma of population, which will probably never happen. ;)

So I'm pretty curious:
-- Are there any major studies done on the "1%" of the most sensitive humans (e.g. people least tolerant to CRTs, people least tolerant to VR, people least tolerant to motion blur, etc)
-- Are there any major studies done on the persistence-versus-motion-blur tradeoff? More specifically any discomfort caused by flicker, versus any discomfort caused by motion blur?

If no studies yet -- these are clearly useful suggested future areas of study for science, since they apply to "displays of the future" technologies, including those covered by Blur Busters.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

ScepticMatt
Posts: 36
Joined: 16 Feb 2014, 14:42

Re: So what refresh rate do I need? [Analysis] [very good on

Post by ScepticMatt » 18 Feb 2014, 11:51

Chief Blur Buster wrote: So I'm pretty curious:
-- Are there any major studies done on the "1%" of the most sensitive humans (e.g. people least tolerant to CRTs, people least tolerant to VR, people least tolerant to motion blur, etc)
-- Are there any major studies done on the persistence-versus-motion-blur tradeoff? More specifically any discomfort caused by flicker, versus any discomfort caused by motion blur?
I found some studies a while ago about flickerand headaches in sensitive humans/migraine/epilepsy, most often in relation to magnetically balanced fluorescent lights and CRTs. Sensitive usually have a lower CFF. Here's a random study about incident rates:
The weekly incidence of headaches among office workers was compared when the offices were lit by fluorescent lighting where the fluorescent tubes were operated by (a) a conventional switch-start circuit with choke ballast providing illumination that pulsated with a modulation depth of 43-49% and a principal frequency component at 100 Hz; (b) an electronic start circuit with choke ballast giving illumination with similar characteristics; (c) an electronic ballast driving the lamps at about 32 kHz and reducing the 100 Hz modulation to less than 7%. In a double-blind cross-over design, the average incidence of headaches and eyestrain was more than halved under high-frequency lighting. The incidence was unaffected by the speed with which the tubes ignited. Headaches tended to decrease with the height of the office above the ground and thus with increasing natural light. Office occupants chose to switch on the high-frequency lighting for 30% longer on average.
ImageImage
Source: http://lrt.sagepub.com/content/21/1/11.abstract

About motion blur, I'm not sure. A quick search only turned up some studies of the headache inducing effects of ghosting in stereo vision.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 18 Feb 2014, 13:52

Thanks for the informative diagrams!
Now in regards to the blur-versus-flicker tradeoff, a very understudied area:
ScepticMatt wrote:About motion blur, I'm not sure. A quick search only turned up some studies of the headache inducing effects of ghosting in stereo vision.
I think the concept of the motion blur versus flicker tradeoff is so new, that not much professional science has been done in this territory yet. Displays used to be too imperfect, small, low-resolution, high-persistence, low-framerates, that the crossover point wasn't statistically siginificant.

But today with Oculus VR and low-persistence LightBoost displays at close desktop view distances -- high-def sharp graphics with faster motion and wider FOV -- there is apparently a crossover being pulled into statistically significant human territory. Where we've got serious blur sensitivities colliding with serious flicker/stroboscopic sensitivities the could potentially exclude VR from being usable for say, 1% or 5% of population -- at least until we successfully achieve low-persistence sample-and-hold (or continuous-motion framerateless displays).

The low-persistence publicity claims (which I agree with) where low persistence (with its flicker/stroboscopic effects) are preferred over motion blur, including Oculus, Valve Software, EIZO, etc:
-- low-persistence solves a lot of nausea problems during VR, because motion blur is often a nausea-inducing effect in the virtual-reality usage scenario. You turn your head, you get motion blur forced upon you (e.g. unable to read signs on virtual walls while turning your head), you get nausea, fix by doing low persistence (e.g. necessitiates flicker/strobing to get millisecond-league at today's refresh rates).
-- It happens less often with LightBoost, but as this is the "LightBoost site" on the Internet, I get many people writing in and saw lots of forum posts over the many months, I see dozens of people writing that LightBoost reduced headaches/nausea for them -- in exactly the same way Oculus and Valve Software is claiming for their low-persistence VR. Such claims come in enough numbers that these can't be dismissed.
-- And we've got EIZO advertising the strobed FDF2405W (professional/commercial version of monitor) as reducing eyestrain during panning maps, so clearly EIZO has noticed the motion blur eyestrain problem already and adds strobing (low persistence) to reduce eyestrain.
-- All of the above combined (And my own personal experience too; less eyestrain during fast motion) is massively enough anecdotal evidence to confirm that motion blur can, indeed, in some cases, create significant discomfort for some in some situations -- that exceeds the discomfort of adding flicker/strobing -- situations that make low persistence the lesser of evil (e.g. flicker being more comfortable than motion blur).

I haven't been able to find much study on the motion blur nausea phenomenon which definitely exists. Likewise, I see people who complain about problems with any impulse-driven technology (e.g. can't stand flicker). It's so statistically significant that it is definitely nonzero (and even if it's just tiny, 1% of 1 million is still 10,000) that science can't ignore this potential factor... I'd be very interested to see new science studies in this direction.

Since there's no economically feasible way to combine low-persistence and completely strobe-free, this is a pick-your-poison effect for a tiny fraction of human population, but possibly statistically significant when trying to make certain new technologies popular (e.g. trying to sell millions of displays). Apple wouldn't ever add flicker to their iPads for example, so we'll be stuck with sample-and-hold motion-blur on tablet computers for a long time, and mobile processors won't be able to do low-persistence sample-and-hold for a long time (low-persistence sample-and-hold requires quadruple digit frame rates).

This is an interesting conundrum that Blur Busters has long considered -- how to satisfy flicker-sensitive needs and blur-sensitive needs simultaneously, making even more percentage of population happy even in the torturous extreme VR usage scenario. (Other than this, you then you start going into other "out-of-sync-with-reality" limiting factors such as vertigo -- gravity and geforce vectors not being consistent with what's seen in VR, etc)

For now, the blur-versus-flicker tradeoff is partially solvable by making it a choice -- e.g. by turning ON/OFF low persistence mode (e.g. turning ON/OFF LightBoost), since it's a relatively simple matter to do with strobe backlights, LED arrays and OLEDs, to drive them flicker-free continuous-light or low persistence pulse-driven (at one strobe per refresh). Economically, satisfying 90%+ of population is probably good enough for now in this stage of technological progress...

Do you know of any scientists/researchers/etc that are about to study display/VR phenomena? Or starting to venture into such territory of studying nausea induced by motion blur (in the modern high-def wide-FOV sharp-graphics era)? If so, tell them about this thread, because it contains some due deligence that makes a good launching pad for such a future study.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 18 Feb 2014, 14:30

...And as a separate study, or rolling into the same study, are the few people who prefer 30fps over 60fps, and people who 24fps movies over HFR 48fps movies. And sometimes, there seems to be a hump factor too -- people who prefer sample-and-hold 30fps (stop-motion effects shows, reducing motion blur nausea) or low-persistence 120fps (zero motion blur) but dislike sample-and-hold 60fps (high enough framerate to be continuous motion, but has motion blur).

I think part of this is because the low-framerate nearly-stop-motion-effect (24fps film, 30fps game) sometimes looks preferable to the motion-blur effect (e.g. sample-and-hold 60fps@60Hz). Some people get less nausea watching 24fps film than HFR 48fps film (which still has motion blur, just less of it, but motion looks less stop-motion and looks more continuous motion). There are some people who prefer 24fps movies, while preferring low persistence/120fps for gaming (not liking the 48fps or 60fps sample-and-hold look), possibly because motion blur is more nausea-inducing than the stop-motion effect of 24fps and the flicker effect of low-persistence. But we don't always know the exact causes of these peoples' modern preferences.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

ScepticMatt
Posts: 36
Joined: 16 Feb 2014, 14:42

Re: So what refresh rate do I need? [Analysis] [very good on

Post by ScepticMatt » 18 Feb 2014, 16:43

Relying on anecdotal evidence is a slippery slope, and should only be used in lieu of more solid evidence. All too often selection bias, preference, placebo or self-fulfilling prophecy ruin the cogency of the results.

Looking a bit more into it, so far inability to accommodate could be the explanation for eye strain and fatigue caused by motion blur.
Here is a study showing the effects of out 'unnatural blur':
Within this zone of comfortable viewing, visual discomfort may still occur to an extent, however, which is likely to be caused by one or more of the following three factors: (1) temporally changing demand of accommodation-vergence linkage, e.g., by fast motion in depth; (2) 3D artifacts resulting from insufficient depth information in the incoming data signal yielding spatial and temporal inconsistencies; and (3) unnatural blur.
http://www.cs.sfu.ca/CourseCentral/820/ ... ort-09.pdf
Any idea what additional research keywords for "motion blur" I could look for?

In a similar sense it well known that vection is a cause of nausea, and higher frame rates and field of view increase the strength of the illusion. So HFR movies and VR games must take care to avoid it.
Last edited by ScepticMatt on 18 Feb 2014, 16:53, edited 2 times in total.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 18 Feb 2014, 17:23

ScepticMatt wrote:Relying on anecdotal evidence is a slippery slope, and should only be used in lieu of more solid evidence. All too often selection bias, preference, placebo or self-fulfilling prophecy ruin the cogency of the results.
Agreed as that case may be, it's still a strong enough indicator that new scientific research needs to be done on these modern display phenomenons, under these new, modern variables.
ScepticMatt wrote:Looking a bit more into it, so far inability to accommodate could be the explanation for eye strain and fatigue caused by screen blur.
Here is a study showing the effects of out of focus screens:
http://www.cs.sfu.ca/CourseCentral/820/ ... ort-09.pdf
Any idea what additional research keywords for "motion blur" I could look for?
For the purposes of searching science papers -- most specifically, in terms of terminology, it is not really "screen blur", but motion blur created by tracking eyes on high persistence displays; "persistence" (used in mainstream media) is essentially a synonym around here for the "sample-and-hold" effect (used in science papers). So essentially, it's 'perceived' motion blur since it only occurs when your eyes are moving across a display (like looking at UFO #1 and UFO #2 at http://www.testufo.com/eyetracking ...)

Industry people are calling it "persistence", but the reused term "persistence" (as applied to strobe backlights and pulsed OLEDs) is more or less a synonym to the "sample-and-hold" and "hold-type" terminology used in several of these older papers. Longer hold, longer persistence, so for the purposes of science, these terms are equal, as the media & the science paper uses different terminology. However, these papers only describe the motion blur reduction benefits, and not the flicker-versus-blur tradeoff.

On Google Scholar, there's a bunch of old science papers (many 2005-ish and older) that contains information on "hold-type displays", "MPRT pursuit camera", "sample-and-hold effect", "MPRT", "moving picture response time" (MPRT), "LCD motion blur", but almost all of them come from the pre-strobe-backlight era. Some of the papers are linked from http://www.blurbusters.com/references while others you can find in Google Scholar. The animation that I designed at http://www.testufo.com/eyetracking also demonstrates this too -- Your eyes don't move like digital stepper motors; your eyes move continuously while tracking moving objects (give or take some eye saccades), but the frames are static. The longer frame is static for, the more opportunity it is blurred across your retina as you're tracking your eyes continuously across a finite-framerate display (much like taking a photo while a camera is moved around, e.g. the sample-and-hold effect is a motion blur equivalence to a camera shutter - e.g. eye tracking on a 60Hz sample-and-hold display, creates the same amount of motion blurring as panning a 1/60sec shutter camera across real-world imagery). There is a close relationship between motion blur and the frame rate on a sample-and-hold display.

On sample-and-hold displays (0ms transition), minimum motion blur is one frame length
...(e.g. 60Hz = 1/60sec = 16.7ms persistence = 16.7ms of motion blur = MPRT 16.7ms)
On strobed displays (squarewave), minimum motion blur is the strobe length
...(e.g. 1/1000sec flash at any refresh rate = 1ms persistence = 1ms of motion blur = MPRT 1ms)

For fast-persistence displays (where GtG is insignificant) and for squarewave impulses (strobe backlights), this perfect equivalence (in milliseconds) occurs:
"persistence" == "MPRT" == "moving picture response time" == "hold time" == "sample-and-hold time"

The media uses "persistence", while science papers often use "sample-and-hold" or "MPRT", but you can consider these two terms equivalent when it comes many examples of modern displays.

Other variables (e.g. slower GtG, phosphor decay, etc) will fuzzy this up, but as displays have become faster and more squarewave, these correlations are immediately observed when playing with fast sample-and-hold displays (e.g. 1ms LCD, OLED) as well as strobe-backlight displays or fast-impulse displays (pixel impulses that resemble squarewave), I haven't yet met a display that behaved this way (under high speed camera) that diverged away from this formula. Strobing isn't always perfectly squarewave, and LCD transitions aren't perfectly squarewave but even non-strobed modern LCD displays are already starting to resemble squarewave pixel transitions (Just view http://www.testufo.com/#test=eyetrackin ... eckerboard on any modern faster 1ms or 2ms 60Hz TN LCD, and you'll see the checkerboard pattern effect of pixel transitions starting to resemble more squarewave. That's because 1-2ms GtG is only a tiny fraction of the 16.7ms refresh cycle. Tracking eyes (or camera) essentially mapping the GtG pixel transition as a motion blur trail, and the sharper the GtG transitions are, the sharper the checkerboard boundaries are in this checkerboard illusion. But when you try to view this TestUFO checkerboard illusion on an older LCD, it becomes more gaussian/sinewave when you view the same animation, such as 10-year old monitors, and stops resembling a near-perfect checkerboard pattern as it does on several modern 120Hz/144Hz monitors). Displays that show a near-perfect checkerboard in this test, tend to also very accurately follow the Blur Busters Law formula. Consequently, I've been very reliably able to predict the amount of a motion blur a display will produce, just by measuring strobe length of a strobe-backlight display. Would love to see more research being done in this direction.

It was only around 2011-2012 when LCDs could compress GtG transitions into a fraction of a refresh cycle to an extent necessary to make pixel transitions (GtG) a far more insignificant factor than persistence. That was the necessary ingredient necessary to make active shutter stereoscopic 3D practical, and highly efficient low-persistence abilities via motion blur reduction strobe backlights. I'm not sure if there are any science papers written since 2011 that clearly covers this, since motion blur of LCDs underwent a huge leap forward in clarity. This would be a new era of displays.

The persistence of a sample-and-hold display is equal to the refresh period -- e.g. (1/Hz)ms -- which means 60Hz LCDs have 16.7ms of persistence. So during 1000 pixels/second motion, you've got 16.7 pixel steps between frames. But your eyes (assuming motionspeeds where eyetracking is still relatively accurate) have continuously moved along the motion vector, by that equal distance; smearing a static frame about 16.7 pixels of motion blur. This motion blur effect form perseffect is clearly witnessed at http://www.testufo.com framerate comparision on a 120Hz monitor. 60fps has half the blurring of 30fps, and 120fps has half the blurring of 60fps. This leads to the simple equation, 1ms of persistence (pixel visibility time) translates to 1 pixel of motion blurring during 1000 pixels/second motion, assuming squarewave persistence (e.g. strobe backlights, sample-and-hold on fast LCDs, rolling-scan OLEDs such as Sony Trimaster), framerate == stroberate == refreshrate, consistent motion, and motionspeed well within eye tracking ability (motionspeeds where minor eye saccades not major blurring factor). I am not currently aware of any science paper that confirms this formula that I have reliably confirmed repeatedly with all major brands of strobe backlights; the motion blurring observed is very consistent with the photodiode oscilloscope strobe length. In the past, most displays were too slow and persistence not squarewave (e.g. phosphor decay, plasma subfields, DLP temporal dither, etc), but with strobe backlight displays, the formula I've discovered (Blur Busters Law) of 1ms persistence = 1 pixel of blurring during tracking 1000 pixels/second motion -- which appears pretty accurate both for human vision and for pursuit camera observations. I've written about this formula at http://www.blurbusters.com/lightboost/10vs50vs100 and hinted upon this in the older http://www.blurbusters.com/faq/60vs120vslb ... These are not real science papers, but articles written for high-end mainstream. However, it would be lovely to see a researcher/scientist take upon studying of this and creating a peer reviewed paper (If any is reading -- obviously, I certainly would be happy to help out -- mark[at]blurbusters.com).

When sitting arm's length away from 24" display, 1080p motion running at 960 pixels/second is a close number to 1000 pixels/second but is a number still divisible by 30, 60 and 120, providing a very convenient test case, a motion speed that is relatively accurately trackable by most human users (eye saccades minor enough that you can still count the number of pixels in the TestUFO alien single-pixel eyes, during ~1ms strobing, when the UFO is moving horizontally at 960 pixels/second -- so at this motionspeed, eye saccades aren't a limiting factor to ability to detecting individual pixels). In this situation, adjusting persistence from 60Hz non-strobed (16.7ms persistence = ~16 pixels blurring at 960pps), 120Hz non-strobed (8.3ms persistence = ~8 pixels blurring at 960pps), LightBoost 100% (2.4ms persistence = ~2 pixels blurring at 960pps), LightBoost 10% (1.4ms persistence = ~1 pixel blurring at 960pps), so adjustable-persistence displays such as LightBoost monitors are excellent potential researcher tools in showing the relationship between persistence and motion blur (1ms persistence = 1 pixel of motion blur during 1000 pixels/second) assuming consistent motion of framerate equalling refreshrate and accurate tracking? The formula applies
to both human eye and to tracking cameras (pursuit camera).

-- Any other papers you've found, based on search terms I've suggested, that are new enough to cover strobe backlight behavior?
-- Any scientific paper on any adjustable-persistence displays?
-- Any scientific paper on fast-response displays where GtG/transition/pixel movement is an insignificant factor and persistence is the dominant factor of motion blur? (e.g. squarewave transitions)
-- Any scientific paper that contains/confirms the simple persistence formula (or a derivative thereof) that I've discovered, 1ms persistence = 1 pixel of motion blur during 1000 pixels/second? (I've begun to call this a Blur Busters Law, due to its reliably repeatable observations on strobe-backlight monitors).
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

ScepticMatt
Posts: 36
Joined: 16 Feb 2014, 14:42

Re: So what refresh rate do I need? [Analysis] [very good on

Post by ScepticMatt » 19 Feb 2014, 06:51

Chief Blur Buster wrote: -- Any other papers you've found, based on search terms I've suggested, that are new enough to cover strobe backlight behavior?
-- Any scientific paper on any adjustable-persistence displays?
-- Any scientific paper on fast-response displays where GtG/transition/pixel movement is an insignificant factor and persistence is the dominant factor of motion blur? (e.g. squarewave transitions)
-- Any scientific paper that contains/confirms the simple persistence formula (or a derivative thereof) that I've discovered, 1ms persistence = 1 pixel of motion blur during 1000 pixels/second? (I've begun to call this a Blur Busters Law, due to its reliably repeatable observations on strobe-backlight monitors).
There are lots of them.
Results from a 2 hour search:

Temporal Properties of Liquid Crystal Displays: Implications for Vision Science Experiments
Published: September 11, 2012
Liquid crystal displays (LCD) are currently replacing the previously dominant cathode ray tubes (CRT) in most vision science applications. While the properties of the CRT technology are widely known among vision scientists, the photometric and temporal properties of LCDs are unfamiliar to many practitioners. We provide the essential theory, present measurements to assess the temporal properties of different LCD panel types, and identify the main determinants of the photometric output. Our measurements demonstrate that the specifications of the manufacturers are insufficient for proper display selection and control for most purposes. Furthermore, we show how several novel display technologies developed to improve fast transitions or the appearance of moving objects may be accompanied by side–effects in some areas of vision research. Finally, we unveil a number of surprising technical deficiencies. The use of LCDs may cause problems in several areas in vision science. Aside from the well–known issue of motion blur, the main problems are the lack of reliable and precise onsets and offsets of displayed stimuli, several undesirable and uncontrolled components of the photometric output, and input lags which make LCDs problematic for real–time applications. As a result, LCDs require extensive individual measurements prior to applications in vision science.
Source: http://www.plosone.org/article/info%3Ad ... 48-Kihara1

LCD motion blur reduction using fir filter banks
Published: 7-10 Nov. 2009
Due to the sample-and-hold nature of liquid crystal display (LCD) image formation, LCDs suffer from motion picture blur. This is especially evident during scenes containing fast motion due to the inherent sample-and-hold nature of LCD image formation. Using models for the human visual system (HVS) we take a signal processing approach to solving this problem by pre-processing the data before it is sent to the display. Whereas previous pre-processing approaches either apply a simple high pass filter or an iterative deconvolution algorithm, this work uses a small collection of efficient linear FIR filters to reduce the amount of perceived motion blur. Specifically, we develop a two-channel non-perfect reconstruction filter bank to reduce the motion dependent low pass effects of the HVS. Perceptual tests indicate that our algorithm reduces the amount of perceived motion blur on LCDs at a lower complexity than the existing deconvolution approach.
Source (requires institutional login): http://ieeexplore.ieee.org/xpl/articleD ... er=5413584

Study on LCD Motion Blur Reduction Based on Blind Signal Processing
Motion blur is one of main challenges for using LCDs in television applications. An effective method for reducing the motion blur by blind signal processing was proposed. In this method, the LCD motion blur images are obtained by the MPRT(Motion Picture Response Time), and the motion vectors are estimated using the cepstral method, then the motion vectors are used in a pre-model for LCD motion blur image, and finally an initialization point spread function(PSF) of blind deconvolution can be constructed. The simulation results show that the proposed blind deconvolution can significantly reduce the visible blurring artifact on LCD.
Source(chinese, requires special login): http://en.cnki.com.cn/Article_en/CJFDTO ... 002003.htm

Visual Annoyance and User Acceptance of LCD Motion-Blur
Published: 17 Mar 2009
It has been recognized for some time now that LCD displays will introduce blur when showing moving objects or moving images. Common motion-blur measurement methods permit to picture the blurred profile of an edge moving with a constant velocity. A normalized blurred edge width is then measured for several gray- to-gray transitions to give a motion-blur score of the display under test. However, these objective measurements are partly based on the behavior of the human visual system and it is an open question how well they correlate with subjective experience of observers. In this study, we develop a subjective experiment in order to assess the annoyance and the acceptance of motion-blur. Results are given and compare with measurements data.
Source:
http://www.researchgate.net/profile/Pat ... b70bed.pdf

Comparisons of motion-blur assessment strategies for newly emergent LCD and backlight driving technologies
Published: October 2008
Abstract— Compared to the conventional cathode-ray-tube TV, the conventional liquid-crystal TV has the shortcoming of motion blur. Motion blur can be characterized by the motion-picture response-time metric (MPRT). The MPRT of a display can be measured directly using a commercial MPRT instrument, but it is expensive in comparison with a photodiode that is used in temporal-response (temporal luminance transition) measurements. An alternative approach is to determine the motion blur indirectly via the temporal point-spread function (PSF), which does not need an accurate tracking mechanism as required for the direct “spatial” measurement techniques. In this paper, the measured motion blur is compared by using both the spatial-tracking-camera approach and the temporal-response approach at various backlight flashing widths. In comparison to other motion-blur studies, this work has two unique advantages: (1) both spatial and temporal information was measured simultaneously and (2) several temporal apertures of the display were used to represent different temporal PSFs. This study shows that the temporal method is an attractive alternative for the MPRT instrument to characterize the LCD's temporal performance.
Source (requires institutional login): http://onlinelibrary.wiley.com/doi/10.1 ... 1/abstract

LCD motion-blur analysis, perception, and reduction using synchronized backlight flashing
Published: February 09, 2006
One of the image quality issues of LC TV is the motion blur. In this paper, the LCD motion blur is modeled using a frequency domain analysis, where the motion of an object causes temporal component in the spatial/temporal spectrum. The combination of display temporal low-pass filtering and eye tracking causes the perception of motion blur. One way to reduce motion blur is to use backlight flashing, where the shorter "on" duration reduces the display temporal aperture function, thus improves the temporal transfer function of the display. The backlight flashing was implemented on a LCD with a backlight system consisting of an array of light emitting diodes (LED). The LED can be flashed on for a short duration after LCD reaches the target level. The effect of motion blur reduction was evaluated both objectively and subjectively. In the objective experiment, the retina image is derived from a sequence of captured images using a high speed camera. The subjective study compares the motion blur to an edge with a simulated edge blur. The comparison of objective and subjective experiments shows a good agreement. Both objective measurement and subjective experiment shows clear improvement in motion blur reduction with synchronized backlight flashing.
Source (requires limited institutional login): http://proceedings.spiedigitallibrary.o ... eid=727895

LCD Motion Blur: Modeling, Analysis and Algorithm
Published: 2010
Abstract—Liquid crystal display (LCD) devices are well known for their slow responses due to the physical limitations of liquid crystals. Therefore, fast moving objects in a scene are often perceived as blurred. This effect is known as the LCD motion blur. In order to reduce LCD motion blur, an accurate LCD model and an efficient deblurring algorithm are needed. However, existing LCD motion blur models are insufficient to reflect the limitation of human eye tracking system. Also, the spatiotemporal equivalence in LCD motion blur models has not been proven directly in the discrete two-dimensional spatial domain, although it is widely used. There are three main contributions of this paper: modeling, analysis and algorithm. First, a comprehensive LCD motion blur model is presented, in which human eye tracking limits are taken into consideration. Second, a complete analysis of spatio-temporal equivalence are provided and verified using real video sequences. Third, an LCD motion blur reduction algorithm is proposed. The proposed algorithm solves an l1-norm regularized leastsquares minimization problem using a subgradient projection method. Numerical results show that the proposed algorithm gives higher PSNR, lower temporal error and lower spatial error than motion compensated inverse filtering (MCIF) and LucyRichardson deconvolution algorithm, which are two state-of-theart LCD deblurring algorithms.
Source: http://www.researchgate.net/publication ... bf3534.pdf

Motion deblurring during pursuit tracking improves spatial-interval acuity.
Published: April 2013
The extent of perceived blur produced by a moving retinal image is less when the image motion occurs during pursuit eye movements compared to fixation. This study examined the effect of this reduced perception of motion blur during pursuit on spatial-interval acuity. Observers judged during pursuit at 4 or 8 deg/s whether the horizontal separation between two stationary lines was larger or smaller than a standard. Three different line separations were tested for each pursuit velocity. Each observer performed these judgments also during fixation, for spatial-interval stimuli that moved with the same mean and standard deviation of speeds as the distribution of eye velocities during pursuit. Spatial-interval acuity was better during pursuit than fixation for small or intermediate line separations. The results indicate that a reduction of perceived motion blur during pursuit eye movements can lead to improved visual performance.
Source (may require login): http://www.ncbi.nlm.nih.gov/pubmed/23402872

Edit:
A preview of some of the most interesting papers appearing in the October 2008 issue of the Journal of the SID.
examples:
Comparisons of motion-blur assessment strategies for newly emergent LCD and backlight driving technologies
Motion-blur evaluation: A comparison of approaches
LCD models to analyze and simulate motion artifacts

Abstracts: http://informationdisplay.org/IDArchive ... eview.aspx

A survey on computational displays: Pushing the boundaries of optics, computation, and perception
(Huge paper covering lots of issues and lots of citations)
http://giga.cps.unizar.es/~diegog/fiche ... Survey.pdf

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: So what refresh rate do I need? [Analysis] [very good on

Post by Chief Blur Buster » 20 Feb 2014, 12:22

That took some time to go through!

Very good information. Although most of them come from before the days of efficient strobe/scanning backlights.

There were a lot of weaknesses in old scanning backlights from the days of many of these scientific papers, including backlight diffusion between adjacent scanning backlight segments (which increases the amount of motion blur relative to the strobe length of a scanning backlight segment). When one segment of a scanning backlight lights up, the light can leak into other parts of the display, lengthening persistence (and increasing the amount of motion blur) as a result. I also wrote about how scanning backlights can be much more inefficient than strobe backlights, and TFTCentral wrote about this too. Also not too many years ago, LCD's responded too slowly to really keep GtG transitions from leaking over multiple frames. This would increase the visibility of strobe crosstalk, interfering with motion clarity. You can see the multiple high speed videos of LCDs that I've created:

This old 2007 LCD is not very friendly to scanning / strobed backlights. The GtG transitions leak too much between refreshes.

phpBB [video]


This newer 2012 LCD is very friendly to scanning / strobed backlights.
The GtG transitions stay largely (in what looks like the territory of ~95-99%+) confined within a refresh cycle:

phpBB [video]


There are situations where an LCD is so slow, it can streak over multiple refreshes, much like this one with 3 different refreshes overlapped (a refresh displaying 06, a refresh displaying 07, and a refresh displaying 08).

Image

Now, when GtG starts to stay confined to their own refresh cycles (e.g. 1ms LCD with a 16.7ms refresh cycle), even real world GTG is largely complete by the end of the 16.7ms for the worst transitions. If you look closely in the high speed video of the 2012 LCD, there's still a very faint amount of leakage of unfinished pixel transitions that is still visible in the subsequent strobe, but it now becomes just a faint crosstalk-like artifact (strobe crosstalk, roughly the same intensity as 3D crosstalk but not wearing 3D glasses). On some LCDs it is so faint it is below the visual noise floor for most visual material (for certain regions of the screen, such as screen center).

On such displays, motion blur is dominantly controlled by the strobe flash length of a full-strobe backlight. There is no weak links (GtG is no longer weak link, scanning backlight diffusion is no longer weak link, LED backlight turns on/off nearly instantly with ~<0.1ms white LED phosphor decay. I've noticed full-strobe-flash backlights manage to reach darn nearly the theoretical maximum efficiency (which is 1ms of flash length equalling 1 pixel of motion blurring during 1000 pixels/second -- you can't get any more efficient than this).

This is information that has really been out in the public only less than the last two years, so I am not sure if any scientific papers covers the theoretical motion clarity efficiencies of strobed/scanning backlights, because this was only successfully achieved after 3D-compatible panels came out and strobe backlights began to be added to them (2011 and newer). No commercial LCDs were able to hit anywhere near the neighborhood of theoreteical maximum motion clarity efficiencies before roughly around 2011-ish when LightBoost came out (it was not till late 2012 when the hack was discovered, and popularized).

So I'm really interested in seeing about new science papers about this new era when LCDs finally began to hit near theoretical maximum motion clarity efficiencies (GtG no longer meaninful part of motion blur calculations; and motion blur completely controlled by strobe lengths and unaffected by GtG/diffusion/other inefficiencies).

All the papers you referenced to are all extremely interesting -- I've seen only some of the ones you found, so you found some new gems (even if from the pre-strobe-backlight-era).
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

Post Reply