Refresh rate / GTG relationship

[KKNDT]

I find a problem while using Blur Reduction on my XL2540, according to Blurblusters's guidance.

I know: Lower refresh=>Longer time per refresh cycle=>More time for GTG before strobing

However, I also notice: Lower refresh usually=>slower GTG.
If you use the same aggressive overdrive to keep the GTG as fast as at higher refresh rate, there will be too much overshoot. And If overshoot accors, it may spend a lot of time for the pixel to return to the target voltage/position. NVIDIA's variable overdrive does a great job keeping a very good balance between response time and overshoot. But they can't keep the response time at the same level accross the whole refresh rate range.

I adjusted everything (overdrive, VT, HZ, area, persistence) but finally end up getting very faint double image during fast moving. It may be acceptable, but not perfect. For the best motion clarity, I still miss my old friend CRT.

240HZ is impressive regarding input lag. But today's panels are far away from perfect in GTG. I think:
1. GTG is not fast enough to perfectlly run at 240HZ. I hope panel manufacturer can improve the response time further before overdrive is used.
2. Response time uniformity is poor. Some transitons take much more time than others, and some slow transitions even exceed the refresh cycle at 240HZ. I have no idea is it possible to use different overdrive impulse to apply on different gray pairs?

[Chief Blur Buster]

However, I also notice: Lower refresh usually=>slower GTG.

There's several variables for GtG:
- The powered momentum (pixel voltage pulse -- fast curve movement at the refresh pass)
- The unpowered momentum (long past voltage pulse -- the continuing curve moment between refresh pass)

Multiple passes can speed up GtG because voltage moment "stairsteps" the GtG closer. That's why often in the ghosting artifact, there's some stairstepping (fainter and fainter ghosts that is not exactly a "Photoshop linear" motion blur, but a stairstepped motion blur). Depends on the LCD, LCD type, and the overdrive used.

Both of this can undershoot or overshoot.

You can see some GtG stairstepping (blur stairstepping) here:





Look closely, you can see 3 stairsteps (ultrafaint ghost, then faint ghost, then main image) corresponding to the voltage-accelerated GtG moments. But the GtG still keeps moving between refresh cycles (pixel transition momentum long after the GtG voltage has passed).

The unpowered GtG momentum doesn't speed up with more frequent refresh cycles, but the powered GtG momentum can speed up with extra refresh cycle passes -- clearing up the GtG stairstepping effect more quickly.

Even OLED can have GtG stairstepping effects -- for example, see http://www.testufo.com/eyetracking on an iPhoneX -- there are faint ghost squares behind the real squares.

The stairstepping is because the actual refresh voltage (which occurs briefly once a refresh cycle) does the fastest GtG movement, while there is still GtG momentum (pixel coasting) between refresh cycles.

Horrendously complicated to make equallized at all refresh rates.

And you need voltage below black, and voltage above white, to reliably overdrive the dim shades and bright shades. But many displays cannot do this, so overdrive is often more ineffective for either dim shades (above black) --or-- bright shades (Below white) --or-- both. For VA panels, it's the dim shades that has big problems. You often need to overshoot the pixel voltage to even just undershoot the pixel, and there ain't always voltage headroom below black, and above white.

Picture this: Overdive accelerating a dark grey pixel to light grey pixel, sometimes means you need to use white's voltage to speed up the transition to light grey. Or using a dark grey's pixel voltage to accelerate a transition from light grey to medium grey. Now, if you're trying to accelerate a dark grey pixel to a full white, you can't always use voltage above white's voltage -- you're stuck not having any overdrive voltage to drive a pixel to full-white. (Or to a full-black, for that matter). It all depends on a specific LCD panel type / tech and how the pixel voltages are set up (zero voltage is black on some displays, and is white on other displays). On some panels, you can often overdrive one end (e.g. white or black) but not the other (e.g. black or white).

So you've got forced asymmetries unless you want to reduce contrast ratio (e.g. 1000:1 down to 800:1). LightBoost intentionally does this (reduce dynamic range -- brighter blacks and dimmer whites) to reduce strobe crosstalk. If you calibrate ULMB this way (brighten your blacks and dim your whites), you will suddenly see strobe crosstalk almost dissappear. For more equallized overdrive with fewer artifacts, slightly narrow your dynamic range to approximately 80-90% of original contrast to reduce a lot of ghosting artifacts. Not everyone is willing to give up dynamic range (contrasty colors) in order to gain more symmetric overdrive speed (fewer ghosting effects) -- it just looks pale and washed out -- but this was necessary for 3D glasses operation when you needed to rapidly switch between 2 frames with no after-effects of the previous refresh cycle (3D glasses crosstalk).

[yehaw]

I find a problem while using Blur Reduction on my XL2540, according to Blurblusters's guidance.

I know: Lower refresh=>Longer time per refresh cycle=>More time for GTG before strobing

However, I also notice: Lower refresh usually=>slower GTG.
If you use the same aggressive overdrive to keep the GTG as fast as at higher refresh rate, there will be too much overshoot. And If overshoot accors, it may spend a lot of time for the pixel to return to the target voltage/position. NVIDIA's variable overdrive does a great job keeping a very good balance between response time and overshoot. But they can't keep the response time at the same level accross the whole refresh rate range.

I adjusted everything (overdrive, VT, HZ, area, persistence) but finally end up getting very faint double image during fast moving. It may be acceptable, but not perfect. For the best motion clarity, I still miss my old friend CRT.

240HZ is impressive regarding input lag. But today's panels are far away from perfect in GTG. I think:
1. GTG is not fast enough to perfectlly run at 240HZ. I hope panel manufacturer can improve the response time further before overdrive is used.
2. Response time uniformity is poor. Some transitons take much more time than others, and some slow transitions even exceed the refresh cycle at 240HZ. I have no idea is it possible to use different overdrive impulse to apply on different gray pairs?

This is a very interesting topic. It could be why I perceived my 240hz display as more blurry at lower refresh rate than my overclocked 165hz display at times.

Have you compared to an older 144hz and experience same issues?

[KKNDT]

This is a very interesting topic. It could be why I perceived my 240hz display as more blurry at lower refresh rate than my overclocked 165hz display at times.

Have you compared to an older 144hz and experience same issues?

Sorry, I don't have a 144HZ display to do a side-by-side comparation. Also, I have returned XL2540 due to quality issue.

Personally, I do not expect nice motion clarity deliverd by today's gaming monitors. So, usually I can live with the motion blur on sample and hold 144/240HZ TN/AHVA. But I do really hate double image effect with strobed backlight.

Here's what I find:
1. For a specified panel, if you use lower refresh rate, you have to get GTG slower, or you tend to receive more overshoot. That's what you can see from NVIDIA's overdrive (like PG258Q) VS aweful overdrive ( like AG251FZ reviewed by TFTcentral)
2. 144HZ AHVA panels have much more uniform GTG transitions accross all the gray pairs than these 144/240HZ TN panels.
3. 240HZ TN panels (XL2540, PG258Q... etc.) don't show very obvious response time advantage over older 144HZ TN panels. Both have approx. 3MS GTG on average.


PG258Q. Good overdrive, keeping the balance between response time and overshoot, but GTG gets slower when you use lower refresh rate

http://www.tftcentral.co.uk/images/asus ... onse_6.png
http://www.tftcentral.co.uk/images/asus ... onse_5.png
http://www.tftcentral.co.uk/images/asus ... onse_4.png


AG251FZ. Aweful overdrive, response time is pleasing, but too much overshoot

http://www.tftcentral.co.uk/images/aoc_ ... onse_7.png
http://www.tftcentral.co.uk/images/aoc_ ... onse_6.png
http://www.tftcentral.co.uk/images/aoc_ ... onse_5.png



PG258Q. Poor uniformity GTG speed on TN panels even with good overdrive

http://www.tftcentral.co.uk/images/asus ... onse_8.png


PG279Q. Very good uniformity GTG on 144HZ AHVA panels

http://www.tftcentral.co.uk/images/asus ... onse_9.png



I need some time to absorb what Chief replied to me.

[Chief Blur Buster]

But I do really hate double image effect with strobed backlight.

The triple lock is numero uno for ULMB perfection (if clarity more important than lag).

Fixing strobing double images as much as possible is done by.

1. VSYNC ON. Frame rate = refresh rate = Strobe rate. The triple lock.
   Otherwise multi-strobe per frame create multi image effects.
   If you hate lag, use HOWTO: Low-Lag VSYNC ON
2. Slightly lower Hz. 100Hz usually has less crosstalk than 120Hz or 144Hz.
   And it is easier to match frame rate = strobe rate *AND* less issue with GtG limitations
   Make sure your GPU always stays matched frame rate = refresh rate = strobe rate. No frame drops!
3. If strobe crosstalk is still a problem, use slightly smaller dynamic range.
   Recalibrate your monitor to brighten your blacks & dim your whites.
   Because it reduces overdrive-related strobe crosstalk by giving overdrive overshoot-control room

And the strobing-amplified microstuttering completely disappear (if you have a good smooth mouse).
That's the way you use a strobe backlight to get the single-image CRT motion clarity of yesteryear gaming.
There might be ultra-faint strobe crosstalk but the annoying intense double-images stop (as long as your monitor isn't the main limiting factor in strobe crosstalk).

Not as good for CS:GO competitions (VSYNC OFF is usually better for that for most people)

But it is great for frantic solo RTS games and fast-scrolling scenery/playfields. Blurfree, stutterfree no-double-image panning, with the game as perfectly smooth as TestUFO. You've seen Super Mario Brothers scrolling on a CRT tube? It's just like that in RTS games for me now, with the triple lock technique. Zero stutter + zero blur. Some monitors even also manage to have strobe crosstalk disappear below human-detectability threshold for the center 1/2 or 2/3 of the screen (except top/bottom screen edges that still has crosstalk due to LCD GtG limitations).

Big problem however, is poor color quality of some monitors' strobing. So if good quality strobing is your priority, choose wisely in addition, above-and-beyond, the above triple-lock technique.

This may not be for you but readers need to know how to use ULMB properly if completely blurfree+stutterfree "motion clarity nirvana" is the priority.