Your post is about what I expected, but is easy for me to explain and is pretty much elementary display science to my head. Hopefully this post has helped (partially) clarify things.MaximilianKohler wrote: ↑26 May 2020, 23:02I'm very confused by this thread and by reading in other places that "a 144Hz monitor 140fps looks better than a 240Hz 140fps".
I'm on the 144hz XG2402 mentioned in the OP. Overdrive (Rampage Response) is ON "Faster". Vsync is OFF.
In Talos Principle:
Freesync OFF, no frame limit, 160-190 FPS = not full tearing, but not smooth.
Freesync OFF + 144 frame rate limit = obvious tearing.
Freesync OFF + 141 frame rate limit = not full tearing but very obvious waves.
Freesync OFF + 120 frame rate limit = better, but not completely smooth.
Freesync OFF + 100 frame rate limit = smooth.
Freesync ON + 141 frame rate limit = smooth.
So what I learn from this testing is that having monitor hz significantly higher than your FPS is preferable, and negates the need for vsync or freesync.
It's a function of stutter harmonics and beat frequencies.
You know audio beat frequencies? Same thing with framerate beat-frequencying against refresh rate.
141fps at 144Hz = 3 stutters per second & 3 tearline roll cycles per second
100fps at 144Hz = ultra-high-frequency stutters that blend into motion blur.
Stutter Beat Frequencies
For VSYNC OFF (to get fps above Hz)
147fps at 144Hz = 3 stutter per second
146fps at 144Hz = 2 stutter per second
145fps at 144Hz = 1 stutter per second
For both VSYNC ON and VSYNC OFF:
143fps at 144Hz = 1 stutter per second
142fps at 144Hz = 2 stutter per second
142fps at 144Hz = 3 stutter per second
Tearing Beat Frequencies
For VSYNC OFF (to get fps above Hz)
147fps at 144Hz = Tearline rolls upwards 3 cycles per second
146fps at 144Hz = Tearline rolls upwards 2 cycles per second
145fps at 144Hz = Tearline rolls upwards 1 cycles per second
144fps at 144Hz = Tearline mostly stationary
143fps at 144Hz = Tearline rolls downwards 1 cycles per second
142fps at 144Hz = Tearline rolls downwards 2 cycles per second
142fps at 144Hz = Tearline rolls downwards 3 cycles per second
Now, you see 144fps cap at 144Hz?
-- Tearline may roll slowly. That's the tiny difference (144Hz may be 143.998Hz)
-- Tearline may vibrate. That's the frame rate capping error (caps aren't perfect frametimes)
Next, Understand Stutters & Motion Blur Is Essentially Same Thing
Firstly, the cause of stutters & common display motion blur are the same thing, especially if you closely watch www.testufo.com/vrr .... Low frame rates vibrate like a slow guitar string. High frame rates vibrate so fast like a blurry high-frequency guitar string. This is the regular frame rate (not erratic stutter).
Also look at how stutters blend into motion blur in this variable-speed motion test:
The humankind invention of a frame rate (using static images to create moving images) often creates side effects such as stutter (at low frame rates, like a slow vibrating guitar string) or motion blur (at high frame rates, like a blurry fast vibrating guitar string).
Your actions of eye-tracking across stationary refresh cycles on most displays (LCD displays which are sample-and-hold), means that the static images are smeared across your retinas -- generating the motion blur that you see.
Now, it's a double edged sword. That motion blur can hide tiny microstutters that are small frametime variations. Reduce the motion blur, and the erratic stutters (beat frequencies, or computer freezes, or diskloading stutters, etc) may become easier to see, depending on variables. Increase the display motion blur, and stutters are harder to see. There are a lot of interacting variables at paly.
Observe that regular stutters are the same thing as motion blur. At low frame rate, it is stutter. At high frame rate, it's motion blur. Another good TestUFO demo that demonstrates the "stutters-blend-into-blur" science is www.testufo.com/vrr (another animation!) ... For more amazing information, read the "stutter vibrations blends to display motion blur" section of Blur Busters Law: The Amazing Journey To Future 1000Hz Displays.
Now, once you've considered the "stutters-to-motionblur" continuum, this starts to make more sense.
"erratic stutter" (basically varying frame visibility times) can manifest itself as the beat frequency of frame rate versus refresh rate. Erratic stutters can be low frequency (visible as additional erratic stutters on the regular-framerate stutter of low frame rates, or the regular-framerate motion blur for higher frame rates).
Erratic stutters can also occur at really high frequencies (blending into additional display motion blur). For example a beat-frequency stutter inside another stutter, can be blended. 100fps at 144Hz doesn't have obvious beat frequencies/harmonics. This also means VSYNC OFF 238fps (non-VRR) may look better at 144Hz than at 240Hz, because the "framerate-near-refreshrate" situation creates an amplified-visibility beat frequency situation.
238-vs-240 = 2 stutters per second
238-vs-144 = no obvious divisors/multiples to create hugely visible beat-frequency stutters
Now if you deviated like 119fps at 240Hz or 359fps at 240Hz, that generates a cyclic stutter too. That still becomes visible again, because it's slightly off -- 119fps at 240Hz shows up as something like a stutter per second (albiet more invisible than 119fps at 120Hz due to reduced refresh cycle granularity)
Also, higher Hz also reduces motion blur (double framerate & double Hz = half motion blur), which can make major frametime variances easier to see. Assuming GtG is fast enough to be an insignificant impediment to a refresh cycle, so GtG needs to keep getting faster for higher refresh rates.
Note: The Blur Busters recommendation is that refresh rate upgrades needs to be geometric. Upgrade by 1.5x to remain human noticeable, preferably at least 2x. The 144Hz-to-360Hz upgrade (2.5x) is more similar to the 60Hz-to-144Hz upgrade (2.4x).
Also, if you turn motion blur reduction on (strobing modes such as ULMB, DyAc, ELMB, PureXP, etc), the microstutter of 100fps at 144Hz might become visible again, since strobing amplifies the visibility of microstutters, since display motion blur hides stutter. Also, if you are testing fluidity using mouse movements, your mouse microstutters may generate an additional error margin, so testing fluidity via keyboard strafing is easier. Or you can enter a custom frame rate at www.testufo.com/framerates-versus ...and observe the microstutter beat frequencies and harmonics that way.
If you reduce motion blur (higher Hz or motion blur reduction), then you do have to eliminate other microstutter error margins (game engine stutter, mouse microstutter, etc). Increasing mouse DPI (1600dpi and reducing sensitivity to 1/4 in-game) can also reduce mouse microstutter error margin. A 400dpi mouse doing 2000 pixels over 1 inch at your preferred sensitivity setting, will go at only 100 positions per second if you move the mouse 1/4th of an inch, creating a mouse-dpi-capped frame rate of 100 frames per second (steppy-steppy-steppy mouseturns). If you lower DPI and juice-up the sensivity, and then try to mouse slowturn, you'll see mouse turns become granular (as if mouseturns were really slow frame rate when done slowly). This can muddy the microstutter beat frequencies / harmonics, 400dpi generating additional error margins in stutter-testing via mouseturns, so if you stutter-test your mouseturns, make sure you use 1600dpi or 3200dpi (at 1/4th sensitivity or 1/8th sensitivity respectively). Then you'll filter-out more of your mouse microstutter error margin away from your framerate-mismatch-Hz microstutters, making it easier to analyze your framerate-mismatch-Hz. Unless you're instead testing keyboard strafing or other perfect-smooth motion.
How tiny or big microstutters are visible? Depends. On a blurry 60Hz display, a 10ms stutter can be made mostly invisible. Now, on a 0.5ms MPRT strobe-backlight, a 1ms stutter can be visible at fast motion speeds. 1ms error at 4000 pixels/sec = 4 pixel stutter jump. Which may be visible if MPRT is low enough (0.5ms = only 2 pixels of motion blur per 4000 pixels/second motion), but definitely hidden meaninglessly by non-strobe 60Hz (16.7ms MPRT).
For example, a sudden 10ms frametime variance is much more noticeable at 240fps (4ms-4ms-4ms-14ms-4ms-4ms-4ms being a 250% stutter-amplitude change during 4 pixels of motion blur per 1000 pixels/sec) than a sudden 10ms frametime variance at 60fps (16.7ms-16.7ms-16.7ms-26.7ms-16.7ms-16.7ms-16.7ms being a 60% stutter-amplitude change during 16.7 pixels of motion blur per 1000 pixels/sec). That diskload microstutter invisible at 60Hz becomes much more visible at 240Hz consequently as a result.
On the other hand, higher Hz can reduce stutter, because of reduced granularity between frametime and refreshtime, especially on fixed-Hz. A missed refresh cycle is a shorter wait to the next refresh cycle, creating smaller divergence between game time and the time photons hit your eyeballs (aka less stutter).
That depends on many variables, and the answer can be yes or no depending on settings or variables.MaximilianKohler wrote: ↑26 May 2020, 23:02I'm very confused by this thread and by reading in other places that "a 144Hz monitor 140fps looks better than a 240Hz 140fps".
Want to read more Blur Busters articles / forum posts, relevant to the science of stutter and motion blur?
- Motion Blur Reduction FAQ (if you've never used a mode like ULMB)
- Blur Busters Law: The Amazing Journey To Future 1000Hz Displays
- Stroboscopic Effect of Finite Frame Rate Displays
- Pixel Response FAQ: GtG versus MPRT
In many situations, the millisecond is unimportant. In other situations, the millisecond becomes human-visible.
Hopefully my post has explained why this is an assumption -- whatever you see at one refresh rate is not necessarily whatever you assume to see at a different refresh rate. By changing the refresh rate, you've changed the variables, which may or may not make specific stutters more visible.MaximilianKohler wrote: ↑26 May 2020, 23:02Not being able to run games at 240+ FPS seems irrelevant. On a 240hz monitor you could simply cap your FPS at whatever is stable (say 150), and you'd have smooth gameplay. What am I missing?
Blur Busters exists because of understanding these milliseconds, an important aspect of eliminating display motion blur, of our Blur Busters name sake. So we are very familiar with this science.