LCD Refresh rates vs response times

[AndrewAPrice]

Try sitting down in front of a 7 MHz Amiga hooked up to an NTSC TV, and tell me it doesn't flicker (or crawl).

Try 50Hz PAL? In the last few decades 100Hz TVs have become the standard for PAL CRT TVs (at least in Australia). But I pity those people in the 1960s-1980s (and some people with cheaper or hand me down sets today) that spend the better part of the day, day after day, month after month, year after year watching their 50Hz TV putting up with flickering because they don't know any better. Yet health specialists recommend at least 75Hz when dealing with CRT monitors for only a few hours.

[Solar]

Try setting an Amiga to an "Interlace" screen mode, and watch the refresh rate drop to 25/30 Hz. That's fun.

[Owen]

The reason why an LCD cannot refresh faster than 60-75Hz is quite simple.

It is a chip. Like all chips, it has a timing specification which it cannot go beyond. Lets assume that the LCD is connected to it's display source (E.G. PC) with a serial bus per channel (I.E. 3 data pins - R, G and B). That means, at 1080p60 it has to operate at a pixel clock rate of 124Mhz (Assuming that, for a 24-bit LCD, it has a 24-bit data bus to the controller chip).

Next consider than an LCD is built with LCOS (Liquid Crystal on Silicon) technology, which is slower than standard silicon. And that they're huge - which means you don't want to build anything too small (Needed, due to propogation delays, to build it fast), because that decreases yield, and let me remind you that you're talking about a 100 by 75cm chip here - not a 10mm² one like a processor - where errors are much more likely.

So yes, it can't go above 60Hz. Thats fine - our eyes can only really detect motion at 24 - 30Hz anyway, which means 60Hz is fast enough. (Yes, our eyes can detect the presence of motion - if it's blatant, e.g, flicker, at rates higher than that, but LCDs don't flicker and so don't trigger that)

[madeofstaples]

If I'm not mistaken anything faster than approx 30Hz is actually not generally noticeable as that is the approximate we'll use refresh rate of the human eye.

I've read that this value for the human eye is more nearly 60Hz, actually.

I think I can answer this one... the human eye doesn't really have what could reasonably be called a "refresh rate." When light does excite a photochemical in the eye, they excite the associated neuron for that cone/rod for about 1/25th of a second. If the (same or similar wavelength of) light continues to excite the photochemical, eventually it becomes depleted and stops exciting the associated neuron.

So if you focus on staring exactly straight ahead, why doesn't you temporarily lose your vision (due to depletion of photochemicals)? Several reasons:

1. Lateral inhibition. When one cone/rod is excited, it inhibits the cones/rods in its vicinity. This adds contrast to our sight and is responsible for some illusions. When an excited cone/rod is depleted of its photochemical, it stops exciting it's associated neuron and thus stops inhibiting the cones/rods in its vicinity. The cones/rods in the vicinity react to light and inhibit the original cone/rod, giving it time to replenish its photochemicals.
2. Visual processes in the brain. Your brain puts together so much of your vision from memory that you'd probably be surprised. You can't see color in your peripheral vision, yet you don't only experience the sight of color in the middle of your visual field--your brain fills in color in your peripheral for you. It does similar work when it knows you are still looking at something but the photochemicals in the eyes are depleted.
3. Unconscious saccades. Unless you really focus on not moving your eyes at all, they naturally preform very quick jumps in various directions to allow light to hit an entirely different set of cones/rods. Your vision does not jump around because of the visual processes in the brain compensating for the saccades. This happens about 3 times a second when you aren't focusing on staring at something.

So while the 25-30ish Hz thing may refer to the length of time a rod/cone excites the associated neuron, the brain and the eyes can conceivably detect much quicker changes. However, the visual information has to propagate through many neurons before you can consciously perceive the image--new visual information may even inhibit neurons farther down the visual pathway from processing older visual information in light (hah, pun) of better visual information. A neuron can propagate an action potential in about 5ms, but the hierarchal structure of the brain makes a lot of new information readily available to some neurons as they are also receiving older information (usually as a feedback--feedback neurons outnumber feedfoward neurons by almost factor of 10!).

Novel visual input must propagate through about 100 neurons at most before we can consciously recognize the object/scene in the image. After you understand what you're looking at, further visual input need only propagate a fraction of that many neurons; at least 5, considering the parts of the brain to which the visual information always travels, but probably closer to 10-20 neurons in total. But these numbers only tell us how long it'll take for us to become conscious of what we are seeing--there are some shortcuts that the brain uses to jump into fight-or-flight mode when your eyes see something dangerous/life-threatening (so you become engaged slightly before you are necessarily consciously aware of what you are seeing).

All in all, it's possible to perceive much quicker-changing information than 30Hz, however, increasing the refresh rate makes the probability that we'll notice artifacts lower. Once you get up to 85Hz, the probability is very low, but some people still perceive a blur (happens when an object moves over too far a distance in too short a period of time--increasing the framerate lets the eye of these people see the object move through the intermediate positions and thus reduces the perceived blur), and at 120Hz it's probably about negligible. Don't they also have a 240Hz TV now? That seems like a little overkill from the perspective of what the human eye will notice.

[AndrewAPrice]

Don't they also have a 240Hz TV now? That seems like a little overkill from the perspective of what the human eye will notice.

240Hz was chosen because it's easily dividable into 24, 30, and 60 and you can display the native content of various formats without having to blend frames.

[Solar]

And because larger numbers sell products.

[AndrewAPrice]

And because larger numbers sell products.

Such as trying to advertise 3,000,000,000Hz processors.

[madeofstaples]

Don't they also have a 240Hz TV now? That seems like a little overkill from the perspective of what the human eye will notice.

240Hz was chosen because it's easily dividable into 24, 30, and 60 and you can display the native content of various formats without having to blend frames.

Hm? 120Hz is dividable into 24, 30, and 60... so I'm guessing the keyword is "easily", but if that's true, why are the factors 4, 8, and 10 "easier" than 2, 4, and 5?

[Owen]

Don't they also have a 240Hz TV now? That seems like a little overkill from the perspective of what the human eye will notice.

240Hz was chosen because it's easily dividable into 24, 30, and 60 and you can display the native content of various formats without having to blend frames.

But not 50Hz. Grr.

[AndrewAPrice]

Hm? 120Hz is dividable into 24, 30, and 60... so I'm guessing the keyword is "easily", but if that's true, why are the factors 4, 8, and 10 "easier" than 2, 4, and 5?

I'm not sure. 240Hz could be part of the '3D ready' craze because that is 120Hz per eye (so you can natively display 24, 30, and 60Hz content per eye). However I've seen 240Hz monitors and TV's that aren't '3D ready'.

[NickJohnson]

And because larger numbers sell products.

Such as trying to advertise 3,000,000,000Hz processors.

My laptop runs at 900,000,000,000,000,000 nHz, has 17,179,869,184 bits of RAM, takes a mere .01 millifortnights to start up, and consumes only .000000015 GJ/s!