Blur Busters Forums

Who you gonna call? The Blur Busters! For Everything Better Than 60Hz™ Skip to content

Why do "Vertical Totals" help Blur Reduction? (ADVANCED FAQ)

Advanced display talk, display hackers, advanced game programmers, scientists, display researchers, display manufacturers, vision researchers. The masters on Blur Busters.

Why do "Vertical Totals" help Blur Reduction? (ADVANCED FAQ)

Postby Chief Blur Buster » 26 Apr 2017, 12:22

Chief Blur Buster wrote:EDIT:
Superseded by new BlurBusters.com article:

Advanced Strobe Crosstalk FAQ


Why does "Large Vertical Totals" Improve quality of Blur Reduction?

This is an Advanced FAQ

NOTE #1. Skip this if you're using "easy" Blur Reduction modes such as ULMB
NOTE #2. This only affects monitors that can do instant mode (unmodified dotclock) straight to the LCD panel. The most popular monitors supporting this FAQ are currently BenQ/Zowie gaming monitors.

You may have heard talk on Blur Busters Forums about the "Large Vertical Totals" trick during adjustments of BENQ/Zowie monitors (and possibly other monitor models). Some users are using a "VT1350" trick with BENQ/Zowie 144Hz gaming monitors. This stands for "Vertical Total of 1350". (Note, some users are even using "VT1500" or "VT2000" on the new 240Hz gaming monitors)

But why does it help Motion Blur Reduction strobe backlights?
Why do Large Vertical Totals improve the quality of Blur Reduction?
Why do Large Vertical Totals reduce/eliminate the double-image effect (strobe crosstalk) during Blur Reduction?

Large Vertical Totals are achieved by loading a Custom Resolution Utility (usually ToastyX CRU) to increase the size of the vertical blanking interval (Vertical Total equals the size of vertical resolution plus the size of vertical blanking interval).

It can improve the quality of Motion Blur Reduction on certain monitors -- most particularly strobe backlights of BENQ Blur Reduction or BENQ DyAc, and a few others. Motion Blur Reduction is accomplished via the use of an optional strobe backlight mode to produce CRT-clarity motion on gaming monitors (e.g. ULMB, LightBoost, BENQ DyAc, etc).

The short answer is large Vertical Totals speeds up painting of refresh cycles onto the LCD (faster scanout per refresh cycle), creating a longer pause between refresh cycles that makes it easier for strobe backlights to cleanly flash a fully-refreshed LCD panel. But exactly how does this work? This Advanced FAQ explains it all in fuller technical detail:

1. First, understand what strobe crosstalk is

Basically, you turn on Blur Reduction and then suddenly see double-image artifacts during maximum framerate. TestUFO motion tests can help greatly here. From the Motion Blur Reduction FAQ, this is what strobe crosstalk looks like:

Image Bad strobe crosstalk (very bad)

Image Average strobe crosstalk (fant)

Image Above-average strobe crosstalk (none)

One factor is caused by LCD pixel response limitations (GtG -- grey to grey -- the transition of an LCD pixel from one color value to a different color value -- which takes time). However, this can be solved on some displays, as several monitors have managed to go almost completely strobe-crosstalk-free. This is successfully accomplished by putting most of LCD GtG response in total darkness, unseen by human eyes (between strobes). This situation requires LCD GtG to be much faster than a single refresh cycle. To see how this is successfully done, read further.

2. Next, strobe crosstalk can affect different parts of the screen

During viewing of a full-screen horizontal panning motion (such as TestUFO Strobe Crosstalk), often strobe crosstalk is worse in certain horizontal bands across the screen. Ideally, it should be clearest in the center of the screen, and worse at the top/bottom edges. However, some monitors can have strobe crosstalk near the middle of the screen.

Image

3. Strobe timing adjustments ("Area") can change location of strobe crosstalk

Changing the timing of strobe phase (the timing of strobe backlight flash relative to refresh cycle) can change the position of the double-image zone.

The common method of "strobe-phase" adjustment ("Area" setting on new BENQ monitors) by end users -- one views a full-screen pattern such as TestUFO Strobe Crosstalk while adjusting.

The strobe-crosstalk double-image zone moves upwards and downwards -- strobe crosstalk affects a different part of the image. The most common situation is to adjust until the screen looks clearest (especially screen center).

Some monitors permit this adjustments, such as most new high-refresh-rate BENQ/Zowie gaming monitors. They now provide a 100% fully user-adjustable strobe phase setting (called "Area", usually via Service Menu) on all their new strobe-capable gaming monitors. Earlier monitors (e.g. XL2720Z) are adjustable via Blur Busters Strobe Utility.

Also, it is important to note certain (unnamed) monitor manufacturers are unfortunately much worse; with non-adjustable strobe crosstalk in the center of the screen at all refresh rates.

Tip: Double Image effects can also be caused by framerates lower than stroberate -- aka double-strobing a frame, either via monitor limitation or via running framerates lower than refresh rate. This is not strobe crosstalk. Other monitor manufacturers double-strobes at certain (usually lower) refresh rates forcing a double-image effect above-and-beyond strobe crosstalk. Strobe crosstalk is different: A double image effect being caused by only one flash per frame -- strobe-crosstalk creates double images because of strobe backlight flashes occuring during incomplete LCD GtG transitions.

4. Next, see this high-speed video of refresh cycles on a gaming monitor

To understand how Large Vertical Totals can reduce strobe crosstalk, it's time to watch a short high-speed-video.

This educational video shows that:
- Refreshes on an LCD panel are painted onto the panel top-to-bottom.
- Most strobe backlights attempt to flash between refresh cycles -- after letting LCDs finish transitioning (that pixel response, LCD GtG!) in total darkness.

This high speed video shows regular LCD operation, and strobe backlight (LightBoost) operation.

phpBB [video]


This is a high speed video of http://www.testufo.com/flicker running in full screen mode. That pattern is chosen because it is much easier to see the LCD refreshing mechanism via high speed video. That 'fuzzy zone' that wipes from the top to bottom, is the LCD GtG zone -- at approximately 1ms pixel response, the blurry LCD GtG zone is about 1/8th the height of a 120Hz (8.3ms) refresh cycle.

5. Next, understand the timing of strobe backlight operation

I recently created this graphic for a different thread and the Electronics Hacking: Creating a Strobe Backlight article.

Image

If the monitor flashes the strobe backlight while the LCD GtG zone is in the middle of the screen, you get strong strobe crosstalk like the above "Double-Image Effect" diagram above.

Ideally, you want to wait for the LCD GtG zone to go below the bottom edge of the screen before flashing the strobe backlight.

However, on many monitors, the time delay between end-of-scanout (previous refresh cycle) and begin-of-scanout (next refresh cycle) is much less than LCD GtG. This means that the LCD GtG zone can wraparound to the top edge of the screen -- the blurry GtG zone can overlap the bottom edge and top edge of the screen!

Naturally, the faster the scanout (e.g. higher refresh rate), the thicker the LCD GtG zone is, because of the pixel response speed (milliseconds) relative to the time of a refresh cycle (milliseconds). As a rule of thumb, 1ms GtG on a 8ms scanout (120Hz monitor), creates an LCD GtG Zone of roughly 1/8th screen height.

Note: Mathematically, GtG is not a perfect complete 1ms -- as GtG is measured from 10% transition to 90% transition. So there can be still be extremely faint after-images (faint strobe crosstalk) leaking well into multiple refresh cycles on some monitors. However, some panels execute GtG extremely accurately, with virtually no human-perceptible after-images in subsequent refresh cycles in most real-world applications. This is technically very difficult to do, but has been achieved on some panels especially when running a reduced dynamic range (reduced contrast) to avoid the more problematic (slower) GtG pairs. As a result, the image above says "Nearly GtG-complete Refreshed Zone" (refreshed areas) above and below the blurry GtG zone (scanout area that wipes downwards)

6. Understand what a Custom Resolution Utility is

A Custom Resolution Utility is built into NVIDIA Control Panel (Custom Resolution) as well as AMD too. However, for creating Large Vertical Total (that properly 100% bypasses GPU scaling), we recommend the use of ToastyX Custom Resolution Utility. It is usually best to use ToastyX instead of the graphics vendor's utility -- as GPU scaling sometimes internally overrides the Vertical Total.

Image

It gives you the equivalent of an airplane cockpit's view of creating a custom computer resolution, and this is the mechanism that allows you to increase the size of the Vertical Total. To learn more about these settings, see GLOSSARY: Custom Resolution Utility Terms.

CAUTION: Monitors may display strange stuff, go out-of-sync, or go black, if you use an excessively large Vertical Total. To save trouble, check the Blur Busters Forums community for recommended Vertical Total settings for a specific monitor.

7. Finally, Large Vertical Totals help you better hide LCD GtG Zone between refresh cycles

But wait! There's a solution to fix that. You accelerate the scanout of the LCD (scan faster) and pause longer between refresh cycles. This gives more time for LCD GtG to finish between refresh cycles, before flashing the strobe backlight.

So how do you do that? Yep! Large Vertical Totals. Many monitors automatically do that internally (e.g. ULMB and LightBoost) but you can manually accelerate the scanout on some monitors (e.g. most recent strobed BENQ gaming monitors) via a Custom Resolution Utility.

The Zowie XL2720 monitor, for example, supports a Vertical Total of 1350 during 120Hz refresh cycle.

  • 1350 equals 1080 + 270. Visible vertical resolution plus the whole vertical blanking interval (including sync/porches).
  • Monitor will spend 1080/1350ths (visible vertical resolution) of a 120Hz refresh cycle scanning-out onto panel.
  • Monitor will spend 270/1350ths (vertical blanking interval) pausing between refresh cycles
  • 1080/1350ths of 1/120sec equals 6.6ms
  • 270/1350ths of 1/120sec equals 1.7ms
  • 6.6ms is 1/150th of a second.
  • VT1350 for 1080p 120Hz forces the monitor to scan-out a 120Hz refresh cycle faster in just ~1/150th of a second ! Instead of closer to 1/120 second!
The accelerated scan-out followed by longer pause between scanouts, is what allows the 1ms TN LCD GtG to (mostly) finish in dark between strobes/between refreshes.

This helps you put the "LCD GtG Zone" more cleanly between refresh cycles.

This means you've precisely adjusted the Vertical Total and strobe timing to have the backlight turned off during LCD scanout and LCD GtG response. The strobe backlight is flashed only on a more fully refreshed LCD panel before the next refresh cycle. Avoiding the strobe crosstalk of flashing during incomplete LCD GtG transitions. The screen flickers 120 times a second, like a 120Hz CRT, with none of the normal LCD motion blur.

The Large Vertical Total, thusly reduces the amount of strobe crosstalk, by making it much easier to calibrate strobe phase -- you can push the horizontal band of double-image-effect off the bottom edge of the screen before it really wraps around back to the top edge of the screen.

If you've seen an old analog TV with a rolling picture, a Large Vertical total is tantamount to increasing the thickness of the black bar between refresh cycles -- seen on a rolling picture of an old analog TV with a mis-adjusted VHOLD knob.

Image

In other words, VT1350 gives you an invisible 270-pixel-tall blanking interval before the strobe crosstalk zones wraps around between the top/bottom edge of the adjacent refresh cycle. If the strobe crosstalk zone is not taller than the blanking interval -- then you're able to completely hide the LCD GtG Zone (Strobe crosstalk zone) between refresh cycles!

As a rule of thumb, the closer you hit the monitor's maximum refresh rate, the worse the strobe crosstalk becomes.
Large Vertical Totals are easier at lower refresh rates, since the monitor often requires Reduced Vertical Totals (Reduced Blanking) at their highest refresh rates (e.g. 144Hz or 240Hz).

Reduced Blanking is extremely bad for strobe crosstalk, so it is recommended to choose a refresh rate lower than the monitor's maximum refresh rate, for best-quality Blur Reduction operation.

Overdrive settings will amplify/reduce strobe crosstalk. The right amount of overdrive helps reduce strobe crosstalk greatly! Make sure you use the most strobe-optimized overdrive setting (BENQ tends to work best with AMA High -- but toggle all the overdrive settings while viewing full-screen motion -- like TestUFO Strobe Crosstalk -- to find what reduces strobe crosstalk the most)

Hope this helps explains strobe crosstalk science/physics & why Large Vertical Totals help!

8. Not all monitors support large vertical totals.

Please note this trick doesn't work on all monitors. It requires the monitor scaler to support real time scanout (instant mode) from the video signal. Not all monitor scalers do.

Supported monitors include most 144Hz+ BENQ/Zowie gaming monitors with Blur Reduction, including Dynamic Acceleration (DyAc), XL2720Z(version 2+), XL2420Z, XL2411Z, XL2420G, XL2735, XL2540, and XL2546.
NOTE: XL2730 scaler does not pass-through Large Vertical Totals

For LightBoost/ULMB monitors -- NVIDIA already automates the operation of Large Vertical Totals in ULMB and LightBoost -- by having the monitor partially buffer a refresh cycle and then outputting an accelerated scan-out to create the longer pauses between refresh cycles. This makes it easier for users, who can just turn ON/OFF the Blur Reduction mode. NVIDIA licenses these technologies to monitor manufacturers as part of GSYNC Monitors. They are very easy and wonderful technologies with easy one-button enable, often with better strobe-optimized overdrive.

However, simply having a monitor be compatible with Large Vertical Totals, and then providing strobe timing and strobe phase settings -- does makes it much easier for monitor engineering. Multiple multiple manufacturers are able to release sufficiently reasonable-quality Blur Reduction that can be then calibrated by end users for better quality. It is less automated, increases user complexity, but often supports more refresh rates (in 1Hz increments) with more user choice.

A possible intermediate step to make this even easier for users, is future calibration software (possibly included with colorimeters) may be able to automatically create compatible Large Vertical Totals and automatically calibrate strobe timing/phase to approximate optimum settings.

---
Did you know? FreeSync / VESA Adaptive-Sync / HDMI 2.1 VRR -- variable refresh rate is achieved via varying the size the blanking interval between refresh cycles. Basically, a variable Vertical Total that changes every refresh cycle.

NOTE: This will soon be converted to an official BlurBusters.com article.
Copyright (C) 2017 by Mark Rejhon of Blur Busters
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter!
User avatar
Chief Blur Buster
Site Admin
 
Posts: 3659
Joined: 05 Dec 2013, 15:44

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Falkentyne » 27 Apr 2017, 01:41

Excellent post, Chief!

I do need to add that certain refresh rates will not play well with certain vertical total values due to some internal scaler stuff. This was first found by Masterotaku; problems that can arise are moving scanlines, frameskipping, pulsating backlight, and so on. Usually just raising or lowering the VT by 1 point, maybe a little more can fix this.

Example: VT 1350 works at 120hz very well. However it frameskips at 100hz quite often; VT 1354 is much more stable at 100hz.
At 85hz (some people may want to use 85hz if they can't maintain 100hz/100 fps with vsync on in some games, but CAN maintain 85 fps), VT 1350 works fine, but only VT 1501 works well in the "high" (1497-1502) range; all others will frameskip extremely often.
At 120hz, VT 1500 is more reliable than VT 1502.
Falkentyne
 
Posts: 2364
Joined: 26 Mar 2014, 07:23

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Dew » 27 Apr 2017, 04:42

Is this VT tweak a waste of time on the ancient VG278H?
Dew
 
Posts: 10
Joined: 07 Sep 2014, 11:19

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby lexlazootin » 27 Apr 2017, 06:02

Dew wrote:Is this VT tweak a waste of time on the ancient VG278H?


Not at all if your monitor supports back light blur reduction and your VT can be modified.

It's not too common but it does work on Benq monitors.
lexlazootin
 
Posts: 931
Joined: 16 Dec 2014, 02:57

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Vega » 27 Apr 2017, 09:17

Very interesting. I have noticed that NVIDIA with ULMB handles cross-talk better than BenQ natively. You have to tweak the VT with BenQ just to get it to ULMB cross-talk minimization, and like you said, at a lower than maximum monitor refresh rate.
Vega
 
Posts: 118
Joined: 18 Dec 2013, 21:33

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Chief Blur Buster » 27 Apr 2017, 22:07

Vega wrote:Very interesting. I have noticed that NVIDIA with ULMB handles cross-talk better than BenQ natively. You have to tweak the VT with BenQ just to get it to ULMB cross-talk minimization, and like you said, at a lower than maximum monitor refresh rate.

Yes.

That said, the best-tweaked BENQ/Zowie (at same refresh rate) has less strobe crosstalk than many ULMB implementations.

NVIDIA generally does a great job with advanced strobe-optimized overdrive (that even includes Y-axis, to make overdrive more aggressive at the bottom edge -- because there's less time to finish LCD GtG before strobe backlight flash).

But I've seen well-calibrated BENQ monitors (proper Vertical Total, proper strobe phase, proper strobe length) -- at same refresh rate, e.g. 100Hz or 120Hz) have less strobe crosstalk in the center 3/4th of the screen than ULMB monitors (even if the bottommost 1/10th of screen may be worse than ULMB -- the very bottom 1 inch of monitor). Getting a BENQ to look this good, often takes an hour or more of manual calibration (VT tweaking, strobe phase tweaking). While with NVIDIA, it's just pushbutton-easy.

What currently sets NVIDIA and BENQ apart is that BENQ lets you strobe at any refresh rate (albiet several of the monitors single-strobes at 60Hz). For example, if you want strobed Quake at 125Hz, you can. Or if you want to play 75fps@75Hz single-strobe. Or if you want to tolerate more strobe crosstalk, go closer to monitor's maximum refresh rate. Or if you want to reduce strobe input lag, then increase the crosstalk at bottom edge in order to reduce input lag for screen centre. BENQ at least gives you that flexibility of this complex adjustment. Etc.

On the BENQ, basically the strobe crosstalk gradient has a different linearity than the ULMB implementations. For example, the LCD GtG zone is slightly thicker, and has a different gradient of crosstalk intensities. There's no Y-axis component in BENQ's simpler most-strobe-optimized overdrive (AMA High) but what's important is the screen centre actually has less strobe crosstalk than ULMB. NVIDIA's strobe overdrive is better, but forces you to a wall garden of 85Hz, 100Hz and 120Hz ULMB with fixed strobe crosstalk intensity.

Sometimes successfully calibrated to have MUCH, MUCH less strobe crosstalk than ULMB (at low refresh rates) -- if you're using VT2000+ at 120Hz on one of the 240Hz BENQs. Certainly the colors might be poorer than ULMB (a 1/240sec scanout at 120Hz, will inherit the color quality of 240Hz, instead of 120Hz -- overkill-league faster scanouts like that are usually poorer color quality with greyer blacks). But at least, you've got the option to be better than ULMB in strobe crosstalk. And on that topic, 1/240sec scanouts at 120Hz means strobe flash can come a bit sooner -- reducing strobe input lag to be less than ULMB too (without a dreaded center-of-screen crosstalk zone).

You can even intentionally adjust strobe phase for less input lag -- e.g. increase crosstalk at the bottom edge (centre still perfectly clear) for even less input lag without being noticeable in most gaming.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter!
User avatar
Chief Blur Buster
Site Admin
 
Posts: 3659
Joined: 05 Dec 2013, 15:44

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Chief Blur Buster » 29 Apr 2017, 10:49

Dew wrote:Is this VT tweak a waste of time on the ancient VG278H?

Vertical Totals is probably waste of time with LightBoost/ULMB....

BUT

There might be faster delivery of frames to the monitor's framebuffer (reducing LightBoost input lag).

It wouldn't overide the fixed strobe phasing or fixed (internally-accelerated) scanout velocity that LightBoost already overrides.

However, Large Vertical Totals in theory accelerates delivery of refresh cycles to the monitor. Instead of a 120Hz refresh cycle taking 100% of a refresh cycle to deliver, you could deliver the 120Hz refresh cycle to the monitor in 1/150th second. That would make it possible for the LightBoost strobe to flash approximately 1.7ms sooner -- reducing input lag by 1.7ms.

I plan to do a highly accurate input lag tests to test the theory that Large Vertical Totals might be able to slightly reduce ULMB/LightBoost lag (even if it doesn't override strobe phase/timing). SMTT-type lag tests isn't precise enough for definitive conclusions on this type of test, since SMTT works poorly with strobing. A lag-differential test on ULMB/LightBoost lag changes of sufficient accuracy can be done with an oscilloscope which I have.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter!
User avatar
Chief Blur Buster
Site Admin
 
Posts: 3659
Joined: 05 Dec 2013, 15:44

Re: Why do "Vertical Totals" help Blur Reduction? (ADVANCED

Postby Chief Blur Buster » 03 May 2017, 21:28

Hello,

I have converted this thread into a Blur Busters article:

Advanced Strobe Crosstalk FAQ
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter!
User avatar
Chief Blur Buster
Site Admin
 
Posts: 3659
Joined: 05 Dec 2013, 15:44


Return to Area 51: Display Science, Research & Engineering

Who is online

Users browsing this forum: No registered users and 1 guest