165 Hz 2560x1440 Curved 27" G-Sync TN - Asus PG27VQ

[Chief Blur Buster]

My biggest takeaway though is for some reason Lightboost has less strobe-cross talk. Not sure why, if it has to do with the overdrive settings etc..

Lightboost is specifically made for 3D, so it makes sense (what doesn't make sense is not using that overdrive setting for ULMB). With some tweaks it's possible to use ULMB in 3D instead of Lightboost. The 3D crosstalk in ULMB is a quite a lot worse. Little overdrive differences in 2D can make great differences in 3D. It's still fun to use with very low custom refresh rates, for testing.

ULMB and LightBoost have different overdrive adjustments, and dynamic ranges (compressed color range) to assist in speeding up LCD pixel transitions differently for different needs.

LightBoost: More aggressive, dynamic range sometimes reduced, creating worse contrast ratio in exchange for less strobe crosstalk -- very important for 3D glasses
poor contrast, usually brighter blacks -- but less strobe crosstalk

ULMB: Less aggressive (more strobe crosstalk) but preserves full dynamic range of unstrobed (more coloreful)
better contrast, usually dimmer blacks -- but slightly more strobe crosstalk

Sometimes the difference is so small (10-20%) that you need to use a colorimeter to realize you're comparing a 750:1 contrast ratio with 850:1 contrast ratio (for a specific screen). Sometimes it is so tiny that it's within the noisefloor of a colorimeter, (e.g. if LightBoost is essentially compressing the dynamic range by only, say, 5%) -- this helps ensure undershoot/overshoot room beyond the dynamic range for easier overdrive compensation (without using voltages beyond spec of an LCD panel).

Programmer/Engineer/Technical Info: To quickly overdrive a pixel from RGB(255,255,255) to RGB(3,3,3), the pixel might slam against the black floor RGB(0,0,0) and get stuck there without rebounding to RGB(3,3,3) quickly -- or trying to voltage it as RGB(0,0,0) (very small difference to RGB(3,3,3)) means you don't transition the pixel nearly as fast. Overdrive overshoot issue amplifying visibility of unwanted ghosting or strobe crosstalk because pixel can't rebound as fast because it slammed at the end of the color range. Now, if you compress dynamic range where you're using, say, RGB(10,10,10) through RGB(240,240,240) of a panel, then the equivalent of RGB(3,3,3) may be RGB(12,12,12) instead. Now you've got 12 clicks of overdrive overshoot room below, rather than just 3 clicks of overshoot. You can even briefly voltage (overdrive) at RGB(0,0,0) to accelerate the movement from RGB(240,240,240) to RGB(12,12,12) and it'll hit nearer that target color without either slamming against the floor/ceiling nor having too little voltage differentials in the overdrive room below black / above white. More accurate GtG is thus achieved by a narrower dynamic range -- meaning you get lower strobe crosstalk during reduced contrast. You can achieve something roughly similar to LightBoost when doing with ULMB by adjusting black level brighter and white level dimmer (via contrast/brightness adjustments, even via NVIDIA Control Panel) -- the strobe crosstalk goes down quite a bit when you digitally adjust to brighter blacks & dimmer whites during ULMB -- but your contrast ratio gets bad / more washed out...

[Vega]

Mark, my measurements for motion clarity are only derived from your sites MPRT test. Although, once you get below 5ms MPRT (such as strobing the backlight) gets much more difficult, IMO requiring instrumentation to measure properly. MPRT is good for measuring "normal" LCD pixel transitions with the naked eye.

As for those 144 Hz strobed 1080P monitors, I'd much rather take the hit down to 120 Hz and gain all the resolution of 1440P.

Otaku, could you tell me what you did to get ULMB to work at 127 Hz on the PG278Q? I wouldn't be surprised if it worked on the PG27VQ.

Right now I'm just interested in possible overclocking the ULMB above 120 Hz and keeping ULMB on at all times and not kicking out of ULMB when exiting a game to desktop.

[Vega]

NM I got it to work at 127 Hz ULMB. Ya baby!

For reference:

Horizontal total (stock): 2720
Vertical total: 1450

Will net you 127 Hz. Wasn't able to get to 128 Hz due to the sync width and front porch taking up that extra last bit. This monitor does not like it when you mess with horizontal width.

[masterotaku]

I used the 60Hz ULMB guide (viewtopic.php?f=4&t=3501) as reference. Using your 120Hz +5VT pixel clock for other refresh rates (changing HT and VT as needed to get that pixel clock or somewhere near it), you get G-Sync+ULMB at any refresh rate that isn't too tight. That's when I reached the 127Hz limit, but black frames made it unuseable with G-Sync+ULMB, so I settled for 125Hz.

There is more crosstalk than at 120Hz, however. I'd only use 125Hz for Quake 3 (or any other game that likes multiples of 62.5fps), for example.

You can also use the 100Hz and 85Hz pixel clocks to use their ULMB settings instead. Not at too high refresh rates because you'll run out of extra VT to reduce.

[Chief Blur Buster]

Mark, my measurements for motion clarity are only derived from your sites MPRT test.

You need to average white->black and black->white, to get an accurate MPRT.
(Hmmm, maybe I can modify the test to make measuring opposite-colors mandatory! )

1. Test this https://www.testufo.com/mprt#background ... und=ffffff
2. Test this https://www.testufo.com/mprt#background ... und=000000
3. Average 1 and 2 for accurate MPRT

Although, once you get below 5ms MPRT (such as strobing the backlight) gets much more difficult, IMO requiring instrumentation to measure properly.

You can measure strobed MPRT via http://www.testufo.com/mprt by using a Checkerboard Size of 2 pixels or 3 pixels. I'm able to measure strobed MPRTs that way.

The smaller the Checkerboard Size in that TestUFO MPRT test, the slower things move, the easier it is to measure MPRT, but the smaller checkerboard size does create more granular MPRT measurements, so you get some rounding errors (more MPRT stepping issues) but it's still accurate to approximately 20%. 1ms MPRT, 1.5ms MPRT, 2ms MPRT.

[Vega]

Ah, never tried that low of pixel settings. Although it looks like the margin of error is a bit higher since the boxes are so small.

It seems like the PG27VQ transitions to sub 1ms strobe at around ~40 ULMB pulse width on this monitor. Full pulse width is around ~2ms.

[Vega]

Just for future reference if anyone searches for ULMB and desktop, I figured out how to keep the monitor in ULMB in game and desktop.

Switch the "Monitor technology" setting in the NVIDIA control panel from "ULMB" to "Fixed Refresh". The ULMB setting keeps ULMB in games but kicks it to normal mode at desktop.

[Chief Blur Buster]

Interesting tip, Vegas!

[tigerman]

Hi, i would like to know if this monitors suffers of the same problems of most of the older asus models, including my 27'' asus PG278Q.

Basically with 3d enabled it looks like scanlined, on nvidia 3d setup the figures are not filled and there are vertical lines all around, there's a picture here on this thread http://3dvision-blog.com/9009-some-3d-v ... n-3d-mode/

would be nice to know is they have solved this issue in this model before wasting 1000€ and get the same problem i already have.

does anyone with 3d glasses can have a look at this ? thanks.

it's possible to see those scanlines even in 2d when you scroll the browser very fast for example, i barely see those lines but it's noticeable.

thanks