Re: Blur reduction and 240Hz monitors
Posted: 26 Nov 2016, 21:35
Very huge ratio of visible:blanking interval, would pull this off.
Example: 3.5 millisecond blanking interval (using a fast 1/288sec LCD scan at 144Hz). As in a 50%:50% visible:blanking ratio, rather than the usual 95%:5% ratio. This would be quite tantamount to a huge Vertical Total (e.g. Vertical Total 2160 = 1080 active rows of visible pixels + 1080 vblank between refresh cycles -- for a 1080p signal), but done internally inside the display, where the LCD is scanned more quickly per refresh, to allow a larger pause between refresh cycles. Metaphorically speaking, this would be like that 1970s analog TV with a rolling picture -- with the VHOLD black bar -- except the VHOLD black bar becomes much thicker, as tall as the preceding/succeeding image -- a massive blanking interval between refresh cycles. This has huge benefits in greatly reducing strobe crosstalk.
This reduces GtG asymmetry between top and bottom edges of screen.
Some displays like EIZO FG2421 achieved fairly good top-to-bottom crosstalk consistency (at least when the screen was warmed up). It actually scanned the panel at something like 1/240sec, during 120Hz mode. So the blanking interval was mondoo huge, big enough to drive the "slow VA 4ms transition" semi-trailer truck through the gapingly-large VBLANK interval.
Someone would need to (1) enable strobing mode, then (2) hardwire the backlight continuously on, and (3) aim a really good high speed camera .... and then we could determine how the LCD is succeeding in lowering strobe crosstalk.
Whenever strobe crosstalk is consistent from top-to-bottom edge of screen, that often means the top-to-bottom scan is being done really quickly. In theory, you could scan faster than the GtG transition -- for VA pixels of 4ms transition time, and a scan that takes only 3.5ms at 1/288sec scan (for 144Hz) you've refreshed the whole screen so quickly that you're triggering the beginnings of the pixel transition (GtG) for the bottom edge of the screen BEFORE the top edge of the screen has completed transitions (GtG). Now if you've got a large blankning interval pause (e.g. at least 4ms -- often critical for VA-based strobed displays, as VA is slower than TN) ... you've got lots of time for all the GtG pixel transitions to complete. The GtG asymmetry of top/bottom edges can be quite much more subdued with much larger blanking intervals.
The bigger the blanking interval (whether internally done like in VA displays, or manually done like in BENQ XL2720Z "Vertical Total"), the easier it is to reduce GtG asymmetry during backlight strobing (less difference between top/center/bottom edges) and also reduce strobe crosstalk -- since it's easier to hide the LCD GtG pixel transitions between refresh cycles and flash the strobe backlight when the pixel transitions are complete. Ideally, you want GtG as small as possible and the blanking interval as long as possible -- for the cleanest possible strobing.
NOTE: For those familiar with XL2720Z "strobe phase adjustment" -- the horizontal strobe crosstalk band -- is much easier to hide inside a larger blanking interval. Meaning, you can move the "double image" band very far off the bottom edge of the screen long before it reappears back at the top edge of the screen. (Strobe crosstalk is massively amplified when the display flashes the blur-reduction backlight while the LCD pixels are still in the middle of their GtG transitions)
TN 240Hz LCDs can probably achieve super-clean look simply by doing this too for 120Hz and 144Hz strobing, too. Assuming the LCD scan is at the full speed (1/240sec) with a large idle period between refresh cycles. For an 8.3ms refresh cycle, 4ms of scan and 4ms of pause (blanking interval. One could even give the LCD 1ms GtG transition a full healthy 3ms to completely settle, before doing a 1ms strobe flash, before beginning the next refresh cycle. This would produce extremely clean refreshes.
Alternatively, a two-pass refresh cycle is a great alternative to a massive blanking interval. Instead of doing a large blanking interval at 50%:50% visible:blanking, you refresh the whole screen twice with the same frame.
Going even further, a 240Hz LCD panel running in 120Hz mode (or 288Hz LCD panel running in 144Hz) could refresh the whole screen twice (The EIZO FG2421 does this) for the same refresh cycle. This helps clean GtG transition imperfections up even further (ghosting, crosstalk, etc). You'd want to strobe at the very end fo the 2nd refresh pass. The two refresh cycles serves to erase as much of the GtG transition artifacts far before the strobe flash. This creates a very clean image. Since some VA displays do the double-pass refresh trick to clean-up their refresh artifacts (GtG/ghosting/etc) -- this may also be why strobe crosstalk is so low for this display.
I highly recommend 240Hz monitor vendors include an ultra-clean strobed 120Hz mode (two-pass refresh + strobe near the end of the 2nd-pass refresh), as it will reduce strobe crosstalk by more than an order of magnitude. In fact, you don't need large blanking intervals for two-pass refresh techniques -- because you can strobe pretty much almost *anytime*, even in the middle of the second pass, since the first pass already did most of the job already, and still have less strobe crosstalk than single-pass refresh methodologies. The second pass is frosting on the cake. Ideal timing of the strobe will still be near the end of the second pass, but just before the first pass of the next refresh cycle, for even fainter strobe crosstalk. The single pass refresh cycle would never be seen by the human eye, as only second-pass refresh cycles would be strobed -- the only purpose of the two-pass refresh cycles is to clean up the GtG artifacts quicker and reduce asymmetry between top/bottom edges as early as possible before the strobe flash of the most perfect-possible-moment of LCD refresh state.
Example: 3.5 millisecond blanking interval (using a fast 1/288sec LCD scan at 144Hz). As in a 50%:50% visible:blanking ratio, rather than the usual 95%:5% ratio. This would be quite tantamount to a huge Vertical Total (e.g. Vertical Total 2160 = 1080 active rows of visible pixels + 1080 vblank between refresh cycles -- for a 1080p signal), but done internally inside the display, where the LCD is scanned more quickly per refresh, to allow a larger pause between refresh cycles. Metaphorically speaking, this would be like that 1970s analog TV with a rolling picture -- with the VHOLD black bar -- except the VHOLD black bar becomes much thicker, as tall as the preceding/succeeding image -- a massive blanking interval between refresh cycles. This has huge benefits in greatly reducing strobe crosstalk.
This reduces GtG asymmetry between top and bottom edges of screen.
Some displays like EIZO FG2421 achieved fairly good top-to-bottom crosstalk consistency (at least when the screen was warmed up). It actually scanned the panel at something like 1/240sec, during 120Hz mode. So the blanking interval was mondoo huge, big enough to drive the "slow VA 4ms transition" semi-trailer truck through the gapingly-large VBLANK interval.
Someone would need to (1) enable strobing mode, then (2) hardwire the backlight continuously on, and (3) aim a really good high speed camera .... and then we could determine how the LCD is succeeding in lowering strobe crosstalk.
Whenever strobe crosstalk is consistent from top-to-bottom edge of screen, that often means the top-to-bottom scan is being done really quickly. In theory, you could scan faster than the GtG transition -- for VA pixels of 4ms transition time, and a scan that takes only 3.5ms at 1/288sec scan (for 144Hz) you've refreshed the whole screen so quickly that you're triggering the beginnings of the pixel transition (GtG) for the bottom edge of the screen BEFORE the top edge of the screen has completed transitions (GtG). Now if you've got a large blankning interval pause (e.g. at least 4ms -- often critical for VA-based strobed displays, as VA is slower than TN) ... you've got lots of time for all the GtG pixel transitions to complete. The GtG asymmetry of top/bottom edges can be quite much more subdued with much larger blanking intervals.
The bigger the blanking interval (whether internally done like in VA displays, or manually done like in BENQ XL2720Z "Vertical Total"), the easier it is to reduce GtG asymmetry during backlight strobing (less difference between top/center/bottom edges) and also reduce strobe crosstalk -- since it's easier to hide the LCD GtG pixel transitions between refresh cycles and flash the strobe backlight when the pixel transitions are complete. Ideally, you want GtG as small as possible and the blanking interval as long as possible -- for the cleanest possible strobing.
NOTE: For those familiar with XL2720Z "strobe phase adjustment" -- the horizontal strobe crosstalk band -- is much easier to hide inside a larger blanking interval. Meaning, you can move the "double image" band very far off the bottom edge of the screen long before it reappears back at the top edge of the screen. (Strobe crosstalk is massively amplified when the display flashes the blur-reduction backlight while the LCD pixels are still in the middle of their GtG transitions)
TN 240Hz LCDs can probably achieve super-clean look simply by doing this too for 120Hz and 144Hz strobing, too. Assuming the LCD scan is at the full speed (1/240sec) with a large idle period between refresh cycles. For an 8.3ms refresh cycle, 4ms of scan and 4ms of pause (blanking interval. One could even give the LCD 1ms GtG transition a full healthy 3ms to completely settle, before doing a 1ms strobe flash, before beginning the next refresh cycle. This would produce extremely clean refreshes.
Alternatively, a two-pass refresh cycle is a great alternative to a massive blanking interval. Instead of doing a large blanking interval at 50%:50% visible:blanking, you refresh the whole screen twice with the same frame.
Going even further, a 240Hz LCD panel running in 120Hz mode (or 288Hz LCD panel running in 144Hz) could refresh the whole screen twice (The EIZO FG2421 does this) for the same refresh cycle. This helps clean GtG transition imperfections up even further (ghosting, crosstalk, etc). You'd want to strobe at the very end fo the 2nd refresh pass. The two refresh cycles serves to erase as much of the GtG transition artifacts far before the strobe flash. This creates a very clean image. Since some VA displays do the double-pass refresh trick to clean-up their refresh artifacts (GtG/ghosting/etc) -- this may also be why strobe crosstalk is so low for this display.
I highly recommend 240Hz monitor vendors include an ultra-clean strobed 120Hz mode (two-pass refresh + strobe near the end of the 2nd-pass refresh), as it will reduce strobe crosstalk by more than an order of magnitude. In fact, you don't need large blanking intervals for two-pass refresh techniques -- because you can strobe pretty much almost *anytime*, even in the middle of the second pass, since the first pass already did most of the job already, and still have less strobe crosstalk than single-pass refresh methodologies. The second pass is frosting on the cake. Ideal timing of the strobe will still be near the end of the second pass, but just before the first pass of the next refresh cycle, for even fainter strobe crosstalk. The single pass refresh cycle would never be seen by the human eye, as only second-pass refresh cycles would be strobed -- the only purpose of the two-pass refresh cycles is to clean up the GtG artifacts quicker and reduce asymmetry between top/bottom edges as early as possible before the strobe flash of the most perfect-possible-moment of LCD refresh state.