6hours old 2411Z V4 BR problem.

Adjusting BENQ Blur Reduction and DyAc (Dynamic Acceleration) including Blur Busters Strobe Utility. Supports most BenQ/Zowie Z-Series monitors (XL2411, XL2420, XL2720, XL2735, XL2540, XL2546)
Falkentyne
Posts: 2793
Joined: 26 Mar 2014, 07:23

Re: 6hours old 2411Z V4 BR problem.

Post by Falkentyne » 17 Jan 2016, 13:45

all refresh rates higher than 128 hz (sometimes 129hz works) require "LCD Reduced", usually a vertical total of 1098.
For lower refresh rates you can use default VT (I know 100hz uses default VT of 1133 and 120hz uses default VT of 1144), then we have our VT tweak to use VT 1500 for low strobe crosstalk+brighter screen.

You cant use VT tweaks higher than 129 hz. Only some Korean monitors (like QX2414) can use normal VT up to 180 hz. We need "LCD Reduced" VT, which is usually VT 1098.

monitor_butt
Posts: 53
Joined: 06 Aug 2014, 20:23

Re: 6hours old 2411Z V4 BR problem.

Post by monitor_butt » 18 Jan 2016, 01:06

Falkentyne wrote:all refresh rates higher than 128 hz (sometimes 129hz works) require "LCD Reduced", usually a vertical total of 1098.
For lower refresh rates you can use default VT (I know 100hz uses default VT of 1133 and 120hz uses default VT of 1144), then we have our VT tweak to use VT 1500 for low strobe crosstalk+brighter screen.

You cant use VT tweaks higher than 129 hz. Only some Korean monitors (like QX2414) can use normal VT up to 180 hz. We need "LCD Reduced" VT, which is usually VT 1098.
Thanks :)

tobiaskarlsen
Posts: 7
Joined: 12 Feb 2016, 04:39

Re: 6hours old 2411Z V4 BR problem.

Post by tobiaskarlsen » 12 Feb 2016, 07:04

@Falkentyne

Sorry for hijacking this post. I've been reading a fair amount of your posts on this and other related forums. I want to start by saying thank you for sharing your knowledge on this subject.

You are really good at explaining and giving detailed explanations ... but could you explain me the following. I'm not quite sure if I understand it correctly.

Monitor: BenQ XL2411Z V004 @ Nvidia GTX980

Given you use the following custom resolution for 120hz (ToastyX pixel clock patched) set in ToastyX CRU:
1920x1080 Active.
front porch 48 pixels 3 line
sync width 32 pixels 5 lines
Horizontal Total: 2080
Vertical Total :1500
Interlaced: unchecked (don't know what that does)
Choose "Custom extention block" » press "OK" » restart64.exe

1) Which of the following options increases/decreases the brightness / response time (ms)?

STROBE PHASE: 043
STROBE DUTY: 007

OR

STROBE PHASE: 044
STROBE DUTY: 006

2) All this makes Blur Busters Strobe Utility unnecessary (could be conflicting) - and I should remove it... right?!

And... Should single strobe be on or off? And what about Vsync - how does that play in to all of this?

Thank you in advance! :)

Falkentyne
Posts: 2793
Joined: 26 Mar 2014, 07:23

Re: 6hours old 2411Z V4 BR problem.

Post by Falkentyne » 12 Feb 2016, 13:05

Hi. This is going to be a bit long.

And what am I going to say will make the most sense to you if you have this test running so you can see visually what I am talking about.
Also keep the service menu open as well to guide you.

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


You're using a VT Tweak in that example, so the maximum strobe phase allowed here will be 49.7.
The higher the strobe phase, the LOWER the maximum strobe duty will be (the lower the persistence).

This is easy to figure out if you know the math and math isn't hard.

For all monitors (not just Benq), refresh rate persistence (input lag, response time, whatever you want to call it) is:
1000 / refresh rate, answer is in milliseconds.
1000 / 60 = 16.7 ms
1000 / 100= 10 ms
1000 / 120= 8.3ms
1000 / 144= 6.9 ms

Maximum strobe phase = (Refresh rate persistence / minimum strobe persistence).

by default, the minimum strobe persistence depends on the refresh rate, and is equal to the refresh rate persistence divided by 100 (yay more divisions).

Minimum strobe persistence is a strobe duty of 001. It's the base persistence for strobing, and the total persistence is equal to base persistence * strobe duty value.

So again assuming default operation (no VT tweaks),
16.7 ms / 100 = 0.167 (base persistence for 60hz)
10 ms / 100 = 0.1 ms (base persistence for 100hz)
8.3ms / 100 = 0.083 ms (base persistence for 120hz)
6.9ms / 100 = 0.069ms (base persistence for 144hz).

As you can clearly see, if you plug in (maximum strobe phase = refresh rate persistence / minimum strobe persistence),
you get "100" every time. And sure enough in the Benq service menu, the maximum strobe phase is 100. It all adds up.

However, when you use a Vertical Total tweak, when the VT is being run out of specification, the monitor reverts to 60hz pulse widths, thus 0.167ms base persistence, even at higher refresh rates. Why does this happen? I don't know. It just does it. So instead of the pulse width (base persistence) depending on the refresh rate, it becomes a FIXED value of 0.167.

As you can see by plugging in the numbers again, this LIMITS the maximum strobe phase. So : for 120hz: 49.7 = (8.3 divided by 16.7).

This is where everyone gets confused.
NORMALLY, when the maximum strobe phase is 100, strobe duty's valid ranges can be set from 1 to 30 (base persistence * 001 through base persistence * 030). Even when strobe phase is 100. There seems to be a buffer of some sort that allows this.

However, when the maximum strobe phase is limited to a LOWER value, this causes the maximum strobe DUTY to also be limited. Meaning as you raise the strobe phase, the persistence also gets lowered (the screen gets dimmer). I believe this is because you are strobing in the current frame instead of the previous frame, and the "buffer" that existed to allow a strobe duty of 1-30 at a full strobe phase of 0-100 is removed. I believe this buffer why strobe phase starts at "000" instead of "001", because 0-100 is 101 values.

So in your example, Strobe phase 043+duty 007 will be BRIGHTER by 0.167 ms of persistence than strobe phase 044 and Strobe duty of 006.

the response time will be the same.
However there will be MORE crosstalk at the BOTTOM of the screen by a few pixel lines, with a strobe phase of 043 than strobe phase of 044.

Remember what crosstalk is?
Crosstalk is the mixing of the frame data from the next frame with the frame data of the current frame at the same time, on the same pixels. With a "high" strobe phase, the pixels towards the top of the screen will be the "current" rendered frame while the pixels towards the bottom of the screen will be the *NEXT* rendered frme. Since "Next" = later, it's obvious (since monitors refresh top to bottom) that there's more input lag at the bottom of the screen than the top. I hope this is self explanatory.

As you LOWER the strobe phase, you are "moving" the next rendered frame higher up the screen and pushing the current rendered frame up, making that area smaller, and then increasing the crosstalk (which will be that horribly blurry field separating the two frames). As your drop the strobe phase down to 000, the next rendered frame now covers the entire screen (meaning you are dealing now with 1 frame of higher input lag total) and the current rendered frame is completely pushed off the top of the screen. There is now some new crosstalk at the bottom of the screen, which is the next rendered frame +1 (two frames of input lag).

The reason why if you go past strobe phase 049, the backlight shuts off, is because you can't decrease input lag of the current rendered frame. There is no such thing as a previous rendered frame that occurs BEFORE the current (true time) rendered frame, and when you increase the strobe phase, the data at the "bottom" of the screen which is the current frame, gets pushed up and covers more of the screen. Since there is no 'previous' rendered frame left, the strobe persistence gets lowered as there's nothing left to strobe, basically. Once you reach strobe phase 049, you're at the absolute limits of the monitor's own response time, which is 0.167 millliseconds. Go any higher and you exceeded the monitor capabilities and the backlight shuts off, because it can't strobe any faster.

tl;dr:
You need to look at the crosstalk at strobe phase 043+ strobe duty 007 vs strobe phase 044 +strobe duty 006, vs strobe phase 045 + strobe duty 005.
As the strobe phase gets higher, the max brightness/persistence is lowered but the crosstalk is lowered also.
Decide on the balance of crosstalk vs brightness. Whatever you find acceptable for image quality in your game is what you should use.

About the strobe utilty?
To be honest, if you are using a VT tweak, the strobe utility caps the maximum strobe phase at 047 anyway, so to be honest, it doesn't matter. the strobe utility is based on people using a VT Tweak at 120hz refresh rate.
However if you are NOT using a VT tweak at all, OR if you are using 100hz with a VT tweak, then a max strobe phase of 047 is a problem for crosstalk. For 100hz, you probably want a strobe phase of 053 (strobe duty 007) or 054 (strobe duty 006) as the backlight shuts off at Strobe phase 060. So yes, once you are experienced with these settings, using the Service Menu is a better idea.

the strobe utility is based firmly on 120hz with a VT tweak. And yes, "earlier strobe phase" in the windows utility should be "later strobe phase" on the left and "earlier strobe phase" on the right, but that's a different story.

tobiaskarlsen
Posts: 7
Joined: 12 Feb 2016, 04:39

Re: 6hours old 2411Z V4 BR problem.

Post by tobiaskarlsen » 12 Feb 2016, 15:19

Thank you very very much. Now I understand the connection and I can physically see what happens when I change the different things and how they relate :)

But ... Could you be so kind and clarify:

Should "Single strobe" (factory menu) be on or off?

And what about Vsync (on/off) - how does that play in to all of this? does it behave differently / better / worse?

Sorry for all my questions :oops:

tobiaskarlsen
Posts: 7
Joined: 12 Feb 2016, 04:39

Re: 6hours old 2411Z V4 BR problem.

Post by tobiaskarlsen » 12 Feb 2016, 16:06

In UFO Test it looks something like this in the top of the monitor (when running full screen) and in the bottom the trail is opposite (the darker trail to the left and the trail that is less visible to the right)

Is this normal / Is it as good as it gets in relation to ghosting on my monitor)?

Image

Falkentyne
Posts: 2793
Joined: 26 Mar 2014, 07:23

Re: 6hours old 2411Z V4 BR problem.

Post by Falkentyne » 12 Feb 2016, 16:15

Single strobe is for refresh rates lower than 100hz. The monitor will double strobe at 60hz, 75hz and 85hz if single strobe is off. The monitor will single strobe at bad, non adjustable settings with VT tweaks at less than 95hz or at bizarre custom refresh rates (91hz, 80hz), I believe the bad settings it tries to use are strobe duty 020 and strobe phase 040. Turning SS on will force single strobe at all refresh rates and adjustments work.

The Single Strobe option is ignored at 100+hz.

The reason the windows utility written by Chief would forcibly enable single strobe on any slider change was due to a bug in V2 firmware where sometimes after a refresh rate/resolution change at 100-144 hz, the monitor would initialize with improper settings of strobe duty 20 and strobe phase 40 until brightness or any strobe setting were changed, then custom settings would "kick in." This bug was fixed on V3+.

Vsync should be enabled when using blur reduction to avoid stutters or tearing (you should keep FPS=refresh rate for best image quality). While it's ok to have a FPS higher than the refresh rate (this will cause tearing), having a FPS lower than the refresh rate will cause very bad stuttering with strobing amplifying it even more. Vsync enabled causes extra input lag (regardless of what monitor you use) equal to the refresh rate persistence, as you are limited by the frame synchronization to refresh rate.

I suggest you experiment yourself. Your own eyes can give you more compelling results far better than me typing descriptions can.

Falkentyne
Posts: 2793
Joined: 26 Mar 2014, 07:23

Re: 6hours old 2411Z V4 BR problem.

Post by Falkentyne » 12 Feb 2016, 16:20

tobiaskarlsen wrote:In UFO Test it looks something like this in the top of the monitor (when running full screen) and in the bottom the trail is opposite (the darker trail to the left and the trail that is less visible to the right)

Is this normal / Is it as good as it gets in relation to ghosting on my monitor)?

Image
That's crosstalk.
If you see a frame to the 'right' of the UFO, think about the direction the UFO is moving.
If its moving left to right, any objects seen to the RIGHT are what is being rendered "previously" meaning before the current point in time) and anything rendered to the left can be considered "lag" (rendered after this point in time).
This description is "relative", meaning it assumes there is no such thing as the "Current" time.

With a high strobe phase, you will never have crosstalk to the "right" of the current frame. Only to the left (which is the next frame with 1 frame higher input lag of course). BUT you will have MORE crosstalk amount at the bottom of the screen (real life blurry image) with a high strobe phase. That's just how it works. It's a tradeoff.

With a strobe phase of 000, almost the entire screen will have 1 frame higher input lag, but you will have very low crosstalk at the bottom of the screen. But the crosstalk you will "see" is the CURRENT frame at a tiny part of the bottom of the screen while everything else (90% of the screen) is the next frame.

You will see how that works if you change the strobe phase.

You are happy if there is no images being rendered to the right ahead of the primary image (for objects moving left to right).
If part of the image (at the bottom of the screen) has images that seem to be melting in from the "left", that's unavoidable. All crosstalk cant be removed because LCD panels are not fast enough.

If you change the strobe phase from 0-100 (without VT tweak) or 0-49 (with VT tweak @120hz; 50 will shut off the backlight), you should see the positions of the UFO change by a frame to the right or left.

The UFO's moving a frame to the right--in the direction of movement means you are getting "true" input lag (meaning you are getting exactly what is being rendered without extra input lag added). If the UFO's move a frame to the left, you are getting 1 frame HIGHER input lag than normal. Crosstalk is the blending of the current frame and the frame with 1 frame higher input lag (aka the NEXT frame to be rendered).

tobiaskarlsen
Posts: 7
Joined: 12 Feb 2016, 04:39

Re: 6hours old 2411Z V4 BR problem.

Post by tobiaskarlsen » 12 Feb 2016, 16:33

Now I have tried to go back and forth from virtually every conceivable options and see the result on ufotest. Now I understand it :)

Thank you so much!!

Falkentyne
Posts: 2793
Joined: 26 Mar 2014, 07:23

Re: 6hours old 2411Z V4 BR problem.

Post by Falkentyne » 12 Feb 2016, 17:38

Yes it's really simple when you see it in action.

You're probably wondering now what is the difference between Ghosting and Crosstalk.
That's actually rather interesting because they are both related but for completely different reasons.

If there were NO ghosting, there would also be NO crosstalk. Strobed CRT's and strobed/rolling scan OLED's don't have ghosting or crosstalk (CRT phosphor trails isn't ghosting in this sense).

Ghosting is simply pixels transitioning from one color to another slowly and not finishing when the data for the next frame or frames is being displayed. So the current frame's colors get "streaked" and appear as dimmer copies behind it.

Overdrive increases voltages to the pixels to speed up color transitions, however not all colors transition at the same time. So while some colors may transition very fast with overdrive voltage, others may still not transition fast enough while others may "overshoot." Overshoot is basically having too much voltage for that color transition so the colors jump 'past' their mark (think of running past a finish line in a race when you have to stop on it. Maybe a baseball hitter running past the base sandbag is a better analogy, then having to quickly turn back).

The reason why ghosting is the same at the top and bottom of the screen is because pixel transition times are identical at the top and bottom of the screen.

However we still have the problem with refresh times and panels refreshing from top to bottom.
And this is where strobe crosstalk comes in.
it's basically a combination of the same thing which causes ghosting in the first place, combined with how refreshes work (top to bottom), and strobing to have the image flashed on the screen (backlight on) for a very short duration during the refresh cycle and the panel not being fast enough to complete pixel transitions during that 'strobe' flash. So in effect, the very same thing causes both ghosting and crosstalk, but the effect is completely different among both.

Try setting the strobe phase to "000" in testUFO at 120hz again (with VT tweak enabled).

Then DISABLE AMA (overdrive).
Notice the crosstalk "seems" to completely disappear?

Oh the crosstalk is still there, mind you! A large part of it HIDDEN now because the increase in ghosting hides it.
Without AMA, the edges of crosstalk are more hidden under ghosting with only the actual borders of the two "frames" merging being sharply visible, as everything else is just blocked by the ghosting.

With AMA enabled, most of the image clears up because of the overdrive voltage boost to the pixels, but that also (ahem) "clears up" the crosstalk--meaning it becomes MORE visible as it's not hidden under a giant ghost trail, so you see the image persistence from the (previous) frame (remember the bottom or the area "below" the bottom of the screen with strobe phase=000 is what the current frame should be).

Post Reply