Experiments on Adjustable Persistence Displays
LightBoost has long been known to be adjustable in persistence, but only by a factor of about 2x (LightBoost 10% versus 100%). However, I have, here, sitting a BENQ XL2720Z monitor with updated firmware, that is has highly adjustable persistence over nearly a whole order of magnitude
, with the new Blur Busters Strobe Utility:
Known: Persistence is equal to strobe length. (measured by photodiode oscilloscope)
Known: Persistence is equal to sample and hold duration. (measured by photodiode oscilloscope)
Confirmed: 4ms persistence has twice the motion blur of 2ms persistence
Confirmed: 2ms persistence has twice the motion blur of 1ms persistence
Confirmed: 1ms persistence has twice the motion blur of 0.5ms persistence
When I view http://www.testufo.com/photo
at 960pixels/sec (very close to 1000 pixels/second, but still divisible by the 120Hz refresh rate), I am witnessing an easy adjustable-persistence-monitor that helps tests the persistence law of "1ms equalling 1 pixel of motion blur during 1000 pixels/second".
Make sure you're using a stutter free web browser ( http://www.testufo.com/browser.html
) and you are not using Classic mode in Windows (which can interfere with TestUFO). Make sure TestUFO is the only webpage running, and no background applications. Enable BENQ Blur Reduction. Disable all 60Hz monitors (if you're using multimonitor). Now, verifiy TestUFO is running stutter-free at 120 frames per second. At http://www.testufo.com/photo
choose a photo (Quebec, Toronto Map or Alien Invasion). Look at the thin window frames in the "Quebec" animation, look at the street name labels in the "Toronto Map" animation, or look at the eyes/pixel details in the "Alien Invasion" animation. While looking at these fine details, adjust.
At 960 pixels/second, I visually observe:
4ms persistence = ~4 pixels of motion blur during 960 pixels/second
2ms persistence = ~2 pixels of motion blur during 960 pixels/second
1ms persistence = ~1 pixels of motion blur during 960 pixels/second
0.5ms persistence = I have some difficulty telling the difference to 1ms
At 1920 pixels/second, I start to notice benefits of 0.5ms persistence:
4ms persistence = ~8 pixels of motion blur during 1920 pixels/second
2ms persistence = ~4 pixels of motion blur during 1920 pixels/second
1ms persistence = ~2 pixels of motion blur during 1920 pixels/second
0.5ms persistence = ~1 pixels of motion blur during 1920 pixels/second
The motionspeeds of 1920 pixels/second (one screenwidth panning per second) is very common in FPS gaming, during mouse turning/strafing/panning. So, the benefits of 0.5ms persistence is just about visible to my eyes during situations where I'm running at 120fps @ 120Hz, especially when combined with an ultra-smooth 1000Hz mouse (500Hz and less is too microstuttery; see next post below).
The benefits of 1ms persistence and 0.5ms persistence was not noticeable to my eyes at 960 pixels/second or less. I did, however, notice a rather interesting phenomenon: the scrolling screendoor effect (where the LCD pixel grid appear to "scroll" along with the motion) only showed up during 0.5ms strobe flash length, not during 1ms strobe flash length. This makes sense, since the shorter 0.5ms strobes is now sufficiently short to eliminate most of eye-tracking-based blurring of the LCD subpixels during 960 pixels/second - so the screendoor effect now stroboscopically 'scrolls along'.
Eye tracking 1920 pixels/second can still be prone to inaccuracies from saccade effects, but I am definitely seeing human visible differences, with my eyes, between 1ms persistence and 0.5ms persistence (not placebo). It's pretty neat to see further visual confirmation of the formula (which I now call "Blur Busters Law of Persistence").Conclusion: 1ms of persistence (strobe flash length) is equal to 1ms of tracking-based motion blur during 1000pixels/second framerate-refreshrate synchronized motion.