spacediver wrote:Yea, I get the scanning vs strobing backlight distinction, I just don't follow the upper bound on persistence thing. Wouldn't the 3.542 ms persistence calculation be the same under both regimes?
You're right -- as long as the visibility of a pixel is 3.542ms under either regime, the size of the motion blur trail looks identical if the pixel was part of global strobing or scanned strobing. (I'm excluding other non-blur artifacts such as skewing or strobe crosstalk, that behaves differently with global versus scanned strobing)
flood wrote:since each pixel can be controlled independently, it's not necessary to have global strobing like with lcd monitors, so there's no upper bound on the persistence.
Persistence is definitely NOT
related to whether strobing is done globally or scanned one pixel at a time. It doesn't matter from a human eye perspective.
Other side effects may exist such as skewing effects caused by scanning (e.g. http://www.testufo.com/blurtrail
when testing on an iPad in portrait versus landscape mode -- the moving line will visibly skew differently depending on how the iPad tablet is rotated. You also see the line skew more on a 60Hz display than on a non-strobed 120Hz display (when readjusting pixel step for same speed). The line skewing of http://www.testufo.com/blurtrail
disappears instantly when you enable strobing.
Global versus scanned strobing has zero, nada, zilch effect on persistence (a.k.a. motion blur visible to the human eye). ZERO
. There are other pros/cons (e.g. scanning can be in theory done laglessly, etc) but since the topic is persistence...
Persistence is more dependant on amount of lumens output per pixel -- the shorter the persistence, the more lumens per pixel.
For LCD global strobing, it just means a brighter backlight for shorter strobe.
For OLED per pixel strobing, it just means a brighter OLED pixel.
Currently, it's easier to get shorter persistences with LCD backlight/edgelight strobing than with active-matrix OLED pixels. Pixel transistor switching time is more of a limiting factor with OLED strobing than with backlights/edgelights at the moment. You can get down to roughly ~0.1ms persistence with strobed LCDs, probably mostly bottlenecked by the speed of white-LED phosphor, but you can get strobe lengths even shorter if you use RGB LED backlights/edgelights.
Regardless, the shorter the strobe, the brighter the strobe needs to be -- doesn't matter if local or global. But it's easier to have a massive lamp focussing into a prism focussing into the edge of an LCD edgelight. Heck, you could heatink and water-cool the LCD edgelight, going RGB LED to bypass the phosphor decay persistence limitation, to push more lumens into it, to achieve shorter strobes (e.g. 1 microsecond or 10 microsecond). The monitor casing could intentionally be designed to have a bulge around a huge monster of an ultra-bright edgelight, if need be, if a manufacturer wanted to massively supercharge the lumen output for shorter strobe lengths. You need twice the brightness for half length strobes. So 100 times brighter for 1/100th persistence of a refresh cycle, 1000 times brighter for 1/1000th persistence of a refresh cycle, etc. At some point, it becomes silly when we're already far beyond diminishing points of returns. But the fact you can super-charge an LCD backlight through rube goldberg tricks like this, is far easier than trying to design an OLED pixel that can strobe super-brightly to compete.
1ms of persistence translates to 1 pixel of motion blurring at 1000 pixels/second (noticed under best case scenario of refresh-framerate-stroberate locked motion) Personally I can tell the motion blur difference of 0.25ms vs 0.5ms, during ultrafast motion tests at ultra-high resolutions, but just very barely. Diminishing points of returns are really pushing it below 0.1ms (give or take, for the VR head-turning use case -- let's imagine a theoretical distant-future 8K OLED VR goggles where such tiny pixels are easily motion-blurred by head turning, to create noticeable imagery sharpness differences --caused by motion blur-- in stationary versus head-turning situations.)
But that said... as long as OLED becomes bright enough, I'd be perfectly happy with 1ms OLED strobing.
(And just live with the slight amount of motion blur of not being able to go below 1ms)
OLED strobing, even with longer persistences of 2-3ms, looks great when done well (see: Oculus OLED rolling scan). I do still see motion blur even with that, given the use-case of fast head-turning creating fast-panning effects that generates eye-tracking-based motion blur
TL;DR: There's no limit how low persistence you can go for either global or scanned low-persistence. You're still running into other limitations that prevents you from shortening persistence easily (e.g. lumens, phosphor decay, transitor switching speeds, etc).