Blur Reduction ON / OFF Help

[samtronswt]

These are my settings for example - 110hz. I'm pretty confident I did patch because the option for 110hz shows up on my display settings when I go to switch it.



Again I appreciate you taking the time to help me out with my issue!

[Falkentyne]

Yes but you didn't answer my question.

Does the monitor OSD show 60hz when you switch to that refresh rate? (your 110hz or 125hz)?

[samtronswt]

Yes but you didn't answer my question.

Does the monitor OSD show 60hz when you switch to that refresh rate? (your 110hz or 125hz)?

Sorry where exactly would I check that?

[Falkentyne]

OSD ->System->Information.

Just to keep this fast, if it reports "60hz" then the VT tweak is working.
Some refresh rates get reported as 60hz even without VT tweaks, however (75hz, 80hz, 91hz are all reported a 60hz even with default VT).

Absolute fastest way to know that your VT tweak is working: create a 1920x1080 custom resolution @ 120hz.
VT 1500.
front porch 48,3
sync width 32,5
HT 2008, VT 1500
120hz
pixel clock 374.40 mhz.

if the OSD reports 60hz then your pixel clock patch is working.

if brightness is too dim, increase the Strobe Duty.

[samtronswt]

So I made a new CRU at 120hz with 1350VT. From what I see it works but for me to get a half decent brightness I'm happy with I have to set persistence to 3ms on the blur busters strobe utility. That's not good is it? Should I try it at 2080 and 1500 VT as you stated or?

I should add when I go to OSD settings on my benq, it reads - Current Resolution: 1920x1080 @ 60hz, thats when I set it to 120hz

[Falkentyne]

That means the VT Tweak is working.
What do you have the brightness set to? it should be at 100 when using MBR.

I can get decent brightness at *1.0ms" persistence with VT 1500.
1.0 ms persistence (at 100hz refresh rate) with VT 1500 with blur reduction on, is about equal in brightness to Brightness 20 with blur reduction OFF.

Anything above 1.7ms persistence WITH a VT tweak and 100 osd brightness is just too bright.
Normal room viewing is comfortable with 120 cd/m2.

Also going above 1.7ms starts reducing the effect of blur reduction and making things blurry. Lightboost 10% is equal to 1.88ms strobe width persistence at 100hz, and 1.4ms at 120hz. 1.88ms is about equal to strobe duty=011 (in the service menu) with a VT tweak active. Note that the persistence values with a VT Tweak use the 60hz values.

That is why the OSD reports 60hz when using a VT tweak.
Masterotaku made a list of the 60hz persistence values in this thread:

http://forums.blurbusters.com/viewtopic.php?f=13&t=560

Note that yes, at 60hz "single strobe=on" those are the persistence values. But when using a VT tweak (1350, 1360, 1500, etc), regardless of the refresh rate, the monitor switches to the 60hz persistence values instead of the actual persistence values for the current refresh rate. Why? I don't know.

When not using a vertical total tweak, if the OSD identifies the refresh rate directly, the persistence values at the refresh rate are equal to:

(1000 divided by refresh rate) divided by 100.
or (1000 / refresh rate) / 100.

Fun fact: the input lag/response time of any refresh rate is equal to 1000 divided by refresh rate in milliseconds. E.g. 1000 divided by 120 is 8.3 miliseconds (=8.3).

if you're curious why the backlight shuts off if the STROBE PHASE gets too high, I have math to explain that too, if you're interested.

[samtronswt]

Brightness is set to 100 when blur reduction is on.

I'll give the 1500VT setting a go and see how far that gets me and report back later tonight! You were talking about 1500 VT for 100 hz, im assuming its okay to run it at 1500VT for 120hz and 110hz, instead of 1502VT?
And brightness 20 with BR off is the desired goal then? I thought the screen would of been brighter.


I'm a last year engineering student, so I'd definitely be able to pick up your math quick! Hit me with that strobe phase getting high explanation, I'm interested haha!

[Falkentyne]

I wrote an incomplete explanation here.

http://www.overclock.net/t/1580000/benq ... -crosstalk

But to go deeper, do you know the relationship between strobe PHASE and strobe DUTY?
There is a reason why when you increase the strobe duty, the crosstalk seems to move down slightly.
Basically it is how strobe duty and phase are related to the strobe itself.

There are 100 values of strobe phase as you can see from the service menu. This value is important.

Knowing that there are 100 values of strobe phase, how is the PERSISTENCE per point of strobe duty calculated?

Simple.
As you know frame lag (frame time, not absolute input lag) or you can call it refresh response time, is equal to 1000 divided by refresh rate, with the answer in milliseconds. We'll use 100hz since that's an easy number later, but for now we'll use 60hz (single strobe enabled).

1000 / 60=16.7ms response time/frame lag.
However what is the persistence for 60hz for each point of strobe duty?

You can see the answer from this chart.
http://display-corner.epfl.ch/index.php/BenQ_XL2411Z
it's 0.167ms. (he rounded it up to .17).
What's the persistence for 120hz?
Frame time for 120hz is 1000 / 120= 8.3ms.
And according to that chart, the persistence per point of strobe duty is 0.08ms. (e.g. raising strobe duty 1 point at a time increases the persistence (strobe width) by 0.167ms (60hz) or 0.083ms (120hz).

What strobe duty actually does is turn on and off the backlight for a certain amount of time each refresh. the amount of TOTAL time the backlight is on **AND** off is the same at each point of strobe duty per strobe, but increasing the strobe duty makes the ON TIME LONGER and the OFF time shorter (thus==brighter screen, more blurry, less blur reduction).

TFTcentral shows that with an oscilloscope here.
http://www.tftcentral.co.uk/articles/motion_blur.htm

However let's get back to that persistence value.

HOW IS THE PERSISTENCE calculated at each refresh rate?
Remember that frame time of 1000 divided by refresh rate?

Divided THAT Answer by 100 and you get the persistence!!

WAIT.......WASN'T THE MAX STROBE PHASE VALUE 100??
......AND YOU DIVIDE THE FRAME TIME BY 100 to get the persistence!

Thus the strobe phase is related to the total strobe TIME PER FRAME (the strobe time is the PERSISTENCE).

that is why it is 100 max.

However, the DEFAULT strobe phase is NOT 000.
It's 100.
A strobe phase of 100 has the lowest input lag.

Reducing the strobe phase from 100 towards 0 cause the strobe to begin LATER in the current FRAME. This seems backwards but if you do this test from test UFO alien invasion WITHOUT a VT Tweak active, you can see how the UFO's react before and after the crosstalk field zone.

http://www.testufo.com/#test=photo&phot ... &height=-1

Notice that as you reduce the strobe phase, the frame seems to go backwards (towards the left) when the image is moving towards the right? thus you LOSE a frame of input lag (input lag gets worse by one FRAME (the persistence value in total strobe phase values) as you go from 100 to 0.

However if you use a Vertical total tweak, something interesting happens.
The backlight switches to the 60hz persistence values instead of the values that refresh rate was designed for.
What happens at 120hz ?
The backlight shuts *OFF* at strobe duty 050! the maximum strobe duty becomes 049 which is exactly HALF of 100. And considering that 120hz refresh rate is double 60hz and 120hz has a frame time half that of 60hz (twice as fast), each point of strobe phase STILL changes the position of the beginning of the strobe cycle by 0.167ms! Yet the frame TIME Is not 0.167ms anymore, its 0.083ms !
As you can see, due to how strobe DUTY And phase are related, going past 049 strobe phase causes the strobe duty (total persistence) to drop to 0.0--the amplitude of that oscilloscope graph becomes nothing. that is the "OFF" cycle lasts forever and there is no on cycle. thus=shut off backlight when a Vertical total tweak is active.

[samtronswt]

Very very interesting indeed. Didn't know any of that before you explained it, it's still a little bit confusing only because I took it all in at once haha, will have to give that a few reads to properly understand the whole concept.

Also, I set my monitor to 120hz @ 1502VT. I turn blur reduction ON and set my brightness to 100, persistence to 1ms and crosstalk to "Earlier Strobe Phase" on the dial on the strobe utility. Is this the maximum amount of brightness I'll be able to get then? Cause gaming on bo3 looks a lot worse at this brightness then it did before.

And finally, is this damaging to the monitor itself?

[Falkentyne]

The ONLY thing that can POTENTIALLY damage the monitor (the backlight WLED's, not the screen or scaler or anything) is running a higher persistence than what the backlight is rated to use at that refresh rate. Keeping in mind that the monitor is rated for 300 cd/m2 brightness at 100 brightness with blur reduction off (maximum).

Since Benq adopted some of the Lightboost hardware for their own blur reduction, both Lightboost and Blur reduction strobed modes (apparently) use the same wire for the strobe signal, and both increase voltage current to the backlight by about a factor of 1.8x. This is to compensate for the strobing loss of brightness so the screen isn't too dark (this is the main problem with ULMB mode on Gsync monitors...ULMB does not increase backlight current at all...this was first noticed directly in the VG248QE upgrade module).

Since the backlight current is increased by 1.8x, the backlight is rated for the maximum persistence for strobe duty=30.
This put the total brightness just slightly below blur reduction off and brightness=100, since this is rather logical...
But this is based on NOT using vertical total tweaks!

You can see from the display-corner link that the persistence values (per point of strobe duty) range from 0.069ms to 2.08ms at 144hz, 0.083-2.49ms at 120hz, 0.1ms to to 3ms at 100hz, and 0.167ms to 5 ms at 60hz.

However remember what I said about VT tweaks?
Using a VT tweak forces the monitor to change to the 60hz backlight pulse widths (0.167ms to 5 ms). The monitor does this whenever it can't identify the refresh rate. The vertical total is what tells the monitor what refresh rate it is using. You can see this by changing the vertical total at 100hz, from the default of 1133, to 1098. 1098 is the default VT for 144hz. If you create a custom 100hz resolution and use 1098VT, the OSD will report 144hz (even though the video card is running at 100hz).

Similarly, 85hz even though it's not listed as default, is actually defined in the firmware itself as an identifiable refresh rate and a VT of 1124 makes the OSD report 85hz. However, 80hz (VT 1122), 75hz (VT 1119) etc are not defined, so if you created an 80hz or 75hz custom refresh rate with default VT, the OSD would fall back to 60hz pulse widths and would report 60hz.

The same thing happens when using VT's higher than normal, e.g. the VT 1350-1360 range, then the VT 1497-1502 range--the monitor can't identify the VT so it reports 60hz.

However it is now using the 60hz pulse widths. And there you start to see a problem.

120hz: 0.083-2.5ms ---120hz is rated for 2.5ms maximum persistence (strobe duty 30).
But when you use a VT tweak, it uses the 60hz persistence values, which are 0.167ms to 5.01ms (strobe duty 30).
So you are exceeding what the backlight is specified for!
Basically anything past strobe duty 015 is going past 2.5ms and thus running the backlight out of specification. This could potentially shorten the lifetime of the LED's, although this hasn't been proven, but you can see it yourself:

Set 120hz, VT 1500, strobe duty 030.
Then turn OFF blur reduction.

Notice the monitor actually gets slightly DIMMER?
Of course you would never use such a blurry persistence anyway but you see the point.

At 100hz, the cap is strobe duty 018 (this does NOT apply if you are not using a VT Tweak of course then the cap is strobe duty 30 as expected...these monitors were not designed to use VT tweaks...we're just exploiting bugs in the firmware basically), because 100hz is rated for 3 ms maximum, and 0.167ms (VT tweak) x 18 is 3 ms. So as long as you don't exceed what the backlight is specified for you're fine.

Fun fact: 50hz does NOT have any single strobe information on the strobe phase at ALL. It DOES have double strobe information (turning off single strobe will make 50hz strobe twice per refresh, at 100hz pulses), but if you enable single strobe, the monitor fails to strobe at all. Backlight current Is STILL increased by 1.8x, and if the current OSD brightness is set too high or the strobe duty is set too high (the monitor is attempting to strobe but the backlight won't turn off) the backlight gets VERY overdriven, becomes excessively bright, and the monitor overvoltage circuit may kick in and cause the screen to hard power cycle itself as if it were unplugged. Unlike using a too high strobe duty, this *CAN and probably *WILL* damage the monitor.



Why does using a VT Tweak lower strobe crosstalk?
Increasing the vertical total extends the blanking interval. This actually causes the scaler to think it is working with a larger vertical resolution than the default one. The strobe signal is based on the blanking interval, so you're basically strobing a larger screen than the physical screen. With a VT of 1340-1370, the scaler resolution is reported as 1600x1280 (the horizontal is either a misreport or completely irrelevant). With a VT of 1370-1400, the scaler resolution is reported as 1792x1344 (the firmware does not like this resolution and displays "out of range" even though you still get an image (whether or not you get frameskipping depnds on the VT value). VT from 1405 to 1480 are all black screens. VT 1481 to VT 1485 are reported as 1920x1440 (notice the vertical keeps climbing?) and once again the scaler does not like that resolution and complains with an out of range.
VT 1486-1496 are black screens.
VT 1497-VT 1502 LOWER the horizontal value while the vertical remains at 1440. Now it reports 1280x1440! Now the scaler accepts this and doesn't complain about out of range anymore. This seems to be the range that Lightboost uses internally via accelerated scanout, as the crosstalk in LIGHTBOOST mode at Lightboost 10% @ 120hz is IDENTICAL TO 120hz benq blur reduction @ VT 1500, strobe phase 000 and strobe duty 009 ! (1.5ms persistence) while Lightboost 10% at 120hz is 1.4ms persistence. The strobe is using a 1440 vertical size to strobe the screen instead of a 1080p size, so crosstalk is reduced by about 25%--the same amount that the vertical internal size is increased.