NVIDIA G-SYNC: Variable Refresh Rate Monitors!

Posted Oct 19, 2013 by

One small step closer to tomorrow’s Holodeck: NVIDIA G-SYNC! It is a technique for refreshing computer monitors at variable intervals (up to a certain limit). Instead of refreshing monitors at exact intervals, the monitor is refreshed when the GPU finishes generating a frame! [Image: AnandTech liveblog of launch event]

Variable refresh rates combines advantages of VSYNC ON (eliminate tearing) with the advantages of VSYNC OFF (low input lag), while virtually eliminating stutters (no mismatch between frame rate and refresh rate).

It is confirmed that this, alone, won’t eliminate motion blur as completely as strobe backlights (e.g. LightBoost), but this is a great step towards eliminating discrete refresh rates (which creates motion blur even at 144Hz. See photos of 60Hz vs 120Hz vs LightBoost), especially as the 144Hz limit is raised in the future, while using nVidia G-SYNC. Hopefully strobe backlight technologies can be combined with G-SYNC in the future — and hopefully already in some upcoming models.

EDIT:
– All G-SYNC monitors include an official strobe backlight mode, better than LightBoost!
– Mark Rejhon has quickly come up with a new method of dynamically blending PWM-free backlight at lower framerates to strobing at higher framerates; see addendum to Electronics Hacking: Creating a Strobe Backlight. This allows combining LightBoost + G-SYNC without creating flicker during lower framerates!

 

About Mark Rejhon

Also known as Chief Blur Buster. Founder of Blur Busters. Inventor of TestUFO. Read more about him on the About Mark page.

39 Comments For “NVIDIA G-SYNC: Variable Refresh Rate Monitors!”

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Sort by:   newest | oldest | most liked
SpartanGR
Member
SpartanGR

Cause i want a monitor for 3d purposes only (i already own Eizo FG2421). Should i buy one now or should i wait for a G-Sync one?

Like0Vote Down 
10 years 4 months ago
SpartanGR
Member
SpartanGR

Hi

Do you have a clue if 3d games will see any benefits by G-Sync in terms of gameplay?

Like0Vote Down 
10 years 4 months ago
Mark Rejhon
Author
Mark Rejhon

G-SYNC and shutter glasses are currently mutually exclusive (at this time).

However, upcoming G-SYNC monitors will likely have a 3D mode built into them, so they are quite multipurpose (G-SYNC mode, strobe mode, and 3D stereoscopic modes).

Like0Vote Down 
10 years 4 months ago
Neo
Member
Neo

One more thing I just remembered to mention. Couldn’t backlight strobe timing itself be used to create variable refresh rates? When viewing a strobe display, the light discharge is essentially the actual frame display. LED timing is measured in nanoseconds so it would seem to difficult to synchronize the beginning of a strobe with the source native frame rate. For example a 3:2 pulldown signal could be sent to a 60Hz display but the strobe would flash at multiples of 24Hz ie simulate multiple-shuttering of film projectors. The the difference in LCD refresh rate would be masked the same way the slow response time is masked by blanking the light.

Like0Vote Down 
10 years 5 months ago
Mark Rejhon
Author
Mark Rejhon

The problem is, for good motion blur reduction, is that you must stick to only ONE strobe per unique refresh, or you get the double-edge artifact. (e.g. double-edge 30fps@60Hz CRT, or triple-edge artifact 60fps@180Hz PWM).

This interesting effect is seen by viewing http://www.testufo.com on a 60Hz CRT (double-image effect on 30fps), as well as on a LCD with low-frequency PWM (dim to 0%), to understand the bad/deleterious effect of doing multiple strobes per refresh does; it ruins motion clarity for video games. Blur Busters is concerned about maximum possible motion clarity during video games, and very few displays achieve this (except LightBoost, which successfully achieves 1.4ms of persistence via a one-flash-per-refresh strobe backlight).

This is because your eyes is continuously moving when tracking moving objects. The strobes are happening at different instants, and puts copies of the images at spatially-different positions in your retinas, as your eyes are continuously tracking moving objects (Tracking moving objects at 1000 pixels/second, at 200 strobes per second, will have duplicate images spaced 5 pixels apart, as an example).

As a general rule of thumb, (strobe rate) / (frame rate) = number of duplicate images.

Like0Vote Down 
10 years 5 months ago
Neo
Member
Neo

Ahhhh… I see. Thanks for your reply. I thought it might not be that bad considering 24 fps film projection had been using double-shuttering to get a 48 Hz flicker the past century.

I guess the edge artifacts are like the color break-up artifacts frame sequential displays have due to saccadic eye movement?

Like0Vote Down 
10 years 5 months ago
Mark Rejhon
Author
Mark Rejhon

Yes, the double-edge artifact in 24fps film has been quite common for a long time, and studios intentionally add motion blur to films to avoid making this too visible/obnoxious. Plus the projector shutter is often a 180 degree shutter, which only reduces motion blur only ever so slightly. Shorter flickers leads to less motion blur, but you can’t eliminate motion blur on low framerates without introducing obnoxious flicker. So we have to use higher framerates, preferably framerates above flicker fusion threshold, then we can eliminate motion blur far more easily — more than two or three orders of magnitude reduction in motion blur is routinely found in low-persistence CRT’s playing game-based framerate=Hz fast panning material (e.g. the butter smooth zero-motion-blur Nintendo panning effect, using common television CRT’s using less than ~0.5ms persistence — compare that to a whopping 42ms of persistence for a 24fps movie).

Today, digital projectors have eliminated the 180 degree shutters (at least when watching in 2D), so the double-image artifact is gone nowadays from modern movie theaters. And there’s slightly more motion blur (just a hair bit) with 24fps digital cinema, than with 24fps film cinema, as no blackout is needed anymore during moving of the physical filmreel frames, to the next frame. 48fps digital cinema reduces motion blur by 50%, but it’s not zero motion blur.

Yes, the image separation artifact of multi-strobing occurs regardless of whether the strobes are same color or the strobes of various colors in rainbows. For non-rainbow strobing, instead of rainbow duplicate images, you got same-color duplicate images.

Like0Vote Down 
10 years 5 months ago
Neo
Member
Neo

Ahhh… I see. Yes. But I bet I’ll come up with more questions. 🙂

Like0Vote Down 
10 years 5 months ago
Neo
Member
Neo

Does introducing a blanking mechanism (like frame advance in film projectors) such as black frame insertion or backlight blanking to hide lcd pixel response, cause the same amount of flicker as short strobe flashing? Would that still be a way to get even a little bit of a benefit for natively slow frame rate content?

Like0Vote Down 
10 years 5 months ago
Mark Rejhon
Author
Mark Rejhon

Yes, black frame effect / mechanical shutter / etc / does visually the same thing a long strobe.
See animation of black frame insertion at testufo.com/blackframes.

That said, most non-backlight-based black frame insertion methods only do a 50%:50% ratio, which isn’t all that good. So typical symmetric black frame insertion (one black frame for every one real frame) is only a 50% reduction in motion blur — because it is equivalent to a very long strobe flash.

Motion blur is proportional to persistence, and therefore, the visible frame length (regardless of whatever mechanism is used to add black period between refreshes)
A 50%:50% ratio of black frame insertion only reduces persistence by 50%.

1:1 black frame = 50%:50% dark:bright = 50% less motion blur
3:1 black frame = 75%:25% dark:bright = 75% less motion blur
9:1 black frame = 90%:10% dark:bright = 90% less motion blur

At 90% less motion blur, what you saw 20 pixels of motion blurring for a specific speed motion, now becomes only 2 pixel of motion blurring (e.g. TestUFO Moving Photo testufo.com/photo becomes about 10x sharper, compared to 60Hz, when LightBoost is enabled, especially at shorter strobe length settings such as strobe brightness 50% or less).

To better understand how dark ratios (strobe lengths) affect motion blur, read these articles:
– 60Hz vs 120Hz vs LightBoost
– LightBoost 10% vs 50% vs 100%

Animation comparision of software-based black frame insertion:
– Black Frames at 1:1 Ratio
– Black Frames at 2:1 Ratio
– Black Frames at 3:1 Ratio

When testing these animations on a non-strobed LCD display, the animation of 3:1 ratio black frame insertion, you will see that the bottommost object has 75% less motion blur than the topmost object. This educational animation is very helpful in understanding how the dark:bright ratios influences motion blur.

You will clearly notice that flicker duty cycle affects flicker detectability. Long dark periods and short bright periods, eliminates more motion blur, but is easier to detect flicker. That’s a disadvantage we have to compensate by using ultra high refresh rates (e.g. 120fps @ 120Hz) so we can take advantage of strobe backlights without the CRT headaches of 60Hz. Most people are okay with 120Hz strobing, but even 120Hz isn’t good enough for some (people who are so sensitive to flicker, that they get headaches under fluorescent lights, staring at a PWM-driven monitor, etc). So that’s why the great-looking strobe-backlight monitors, often run at triple-digit refresh rates.

Now, doing this in software based manner, your shortest possible strobe length is limited by the length of a monitor refresh. The only way to shorten this further is to do it using hardware-based manner (e.g. strobe backlight). So that you can keep flicker rates above flicker fusion threshold — e.g. 120Hz strobed. That way it looks flicker free.

Conceptually and scientifically, strobe backlights are essentially tantamount to hardware-based backlight-based black frame insertion, as they can flash for a sub-frame period. A strobe backlight flash of 1.4ms (LightBoost=10%) gives you a 6.9ms:1.4ms dark:bright ratio (1/120sec = 8.3ms refresh cycle) which reduces motion blur by 85% relative to regular 120Hz. (That’s 92% relative to regular 60Hz).

Software based black frame insertion is not always a practical way to reduce motion blur, because it flickers quite badly as you have to lower your frame rate in order to take advantage of software based black frame insertion. However, it is quite useful for reduced-blur 60fps on 120Hz monitors:
— Running emulators (e.g. http://www.blurbusters.com/mame), works both on strobed 120Hz and non-strobed 120Hz displays. WinUAE also added software black frame insertion, too.
— Getting 60Hz LightBoost out of 120Hz LightBoost, for 60fps videogames. The goal of the black frame simply hides the unwanted extra strobes, so you effectively get 15.3ms:1.4ms dark:bright ratios (out of a total 1/60sec = 16.7ms) for your 60fps games; a massive 92% motion blur reduction for 60fps emulator games. It will flicker quite badly, however. It does make emulator scrolling ultra-crystal-clear, though!
— Without this, LightBoost doesn’t help emulators much because of the double-strobe per emulator refresh. With this, you get full LightBoost effect at 60fps. (though it flickers like a 60Hz CRT too!)

Makes a lot more sense now? 🙂

Like0Vote Down 
10 years 5 months ago
Neo
Member
Neo

While doing some lite research I happened upon this older report about Texas Instruments figuring out a way to reduce flicker with PWM lighting. Basically it treats most-significant-bits (MSB) and least-significant-bits (LSB) differently based on how much they contribute to flicker. This should work with the PWM of LCD backlights, as well as OLEDS.

Frame-Rate Technique Delivers Flicker-Free Motion-Picture Performance
http://electronicdesign.com/displays/frame-rate-technique-delivers-flicker-free-motion-picture-performance

I wonder if strobing could also be combined with local dimming (local strobing?) backlights

Like0Vote Down 
10 years 5 months ago
Mark Rejhon
Author
Mark Rejhon

It’s REALLY hard to eliminate motion blur without using flicker.

A zero-motion-blur display (CRT, LightBoost) typically have ~1ms of persistence or less. However, flicker free displays (even 1ms-2ms LCD’s) have 1/60sec = 16.7ms of persistence at 60Hz, so it is the dominant cause of motion blur (See http://www.testufo.com/eyetracking), as persistence is the dominant cause of motion blur nowadays.

Video games are more demanding than television/movies for motion clarity:
— Sharper graphics; (movies/TV are fuzzier)
— Closer viewing distances;
— Faster motion in video games than movies/video;
— Ever higher resolutions;
— Bigger field of view (e.g. VR goggles);

So 120fps flicker-free isn’t enough; and 240fps flicker-free isn’t enough either;

At this point, we even see motion blur during 1ms of persistence. 1ms of ‘clean’ persistence (0ms GtG) creates 1 pixel of motion blurring for every 1000 pixels/second. That’s the theoretical best-possible case scenario. To do 1ms of persistence, you need frames that are visible for only 1ms — either flash for 1ms, or use that many frames (e.g. 1000 unique frames of 1ms each). I also posted on AVSFORUM, Why We Need 1000fps @ 1000Hz This Century, though there can be other creative technologies to achieve this in a flicker-free manner, such as refreshing the display more frequently only where the human eye is staring at, etc.

Right now, today, the easiest way to do low persistence is via various flicker methods (strobing, CRT, light modulation, scanning, black frame insertion, plasma, etc). There is no easy way to reduce persistence to less than 1ms without flicker (zero flicker under high speed camera).

I’ll be making a copy of this “1000fps” thread in the Blur Busters Forums when it launches; it’s going to be an interesting theoretical topic thread.

Like1Vote Down 
10 years 5 months ago
PoWn3d_0704
Member
PoWn3d_0704

I heard a rumor that you could send in your Asus VG248QE 144hz and get the G-Sync chip installed. The prototypes at PDX LAN this year were simply modified VG248QE’s. Since I have one that would be amazing. After playing CSGO on a GSync monitor I have to say that I can’t wait

Gsync, lightboost, 144hz. I can’t wait.

Also, why doesn’t ToastyX have a 144strobed setting? Why am I stuck at 120?

Like1Vote Down 
10 years 5 months ago
Mark Rejhon
Author
Mark Rejhon

I heard a rumor that you could send in your Asus VG248QE 144hz and get the G-Sync chip installed. The prototypes at PDX LAN this year were simply modified VG248QE’s. Since I have one that would be amazing. After playing CSGO on a GSync monitor I have to say that I can’t wait

Yes — they were announced to be upgradeable. See this article: Upgrade Asus VG248QE with G-SYNC. We will be posting an upgrade walkthrough.

Also, why doesn’t ToastyX have a 144strobed setting? Why am I stuck at 120?

Strobe backlights require an artificially enlarged vertical blanking interval (to give time for LCD refresh to stabilize before the backlight is strobed), and 144Hz only works during Reduced Blanking intervals. See the LightBoost FAQ and High Speed Video of LightBoost.

Like1Vote Down 
10 years 5 months ago
wpDiscuz

Recent Content

Blur Buster Forums

Blur Busters Forums

TestUFO Motion Tests

TestUFO Motion Tests