..... its like frame-rates in video games: once you're over about 60 frames per second, any more is a waste.
....
Wo ho ho, speaking of bs. You think the average person cannot tell the difference between 60 and 100 fps? Do you do much computer gaming?
Wo ho ho, speaking of bs. You think the average person cannot tell the difference between 60 and 100 fps? Do you do much computer gaming?
http://www.mikhailtech.com/articles/editorials/fps/ (http://www.mikhailtech.com/articles/editorials/fps/)
http://www.mikhailtech.com/articles/editorials/fps/ (http://www.mikhailtech.com/articles/editorials/fps/)
That guy is a great example of a common attitude on the internet: "Just because you know nothing about a subject, don't let that stop you from writing a web page about it!"
"The most common "frame rate" on a television set is 24, with ranges that can go from 18-30 (these are approximations)."
LOL! The correct figures are 30 or 25, depending which country you're in. Typically countries with 60Hz power supplies have NTSC running at 30 frames per second (60 fields per second), and countries with 50Hz power supplies have PAL running at 25 frames per second (50 fields per second). There are exceptions (e.g. PAL60).
24fps is possible on US HDTV, but I doubt he was talking about that. The other frame rates he mentions are nonsense, as the frame rate on all standard TV systems is fixed, not variable.
"Current LCDs don't have higher than 30-40hz refresh rates, but they're progressive."
All CRT PC monitors are progressive. It's TV that's interlaced.
etc etc etc
Is there any good research in this field? Intuition suggests that the required frame rate (such that the frame rate itself is no longer an issue) depends on whether your eyes are tracking the motion on the screen, or just letting it pass. If you're tracking it, you need a very high frame rate, and no blur. If you're letting it pass, then motion blur is OK, and you can allow a lower frame rate.
There must be several studies on this subject?
Cheers,
David.
Do you ever see car wheels appearing to rotate backwards?
You certainly see it on TV, but does it also happen in 'real life'?
If so (I think it does but I'm struggling to separate generalised memories of seeing cars moving into TV and real life ), the speed of rotation of the wheels is faster than the human optical and nervous system can detect. This would give some clues.
It only happens in real life under artificial lighting, which gives a strobed effect. Otherwise, the wheels only look like they're going backwards when they really are!
Imagine a spinning wheel with one spoke (or one dot on the outside). If you take a picture (of flash a light) once per revolution, the wheel wil look like it's stationary, because you always see it at the same point in its cycle. If you take a picture (or flash a light) slightly quicker than this, then you'll only see the wheel just before it completes one rotation, and it will look like it's going backwards. If you take pictures slightly more slowly than once per cycle, the wheel will get to move one rotation and a little bit extra between pictures, so it will look like it's going forwards, but slowly!
It's time domain aliasing. Thinking about it as being similar to the kind of aliasing that ocurrs in audio (that's frequency domain aliasing) suggests that there's still a Nyquist limit, and if we figure out where the human limit is, and set the Nyquist limit above that, then all will be fine. BUT you need an anti-alias filter before converting a real scene into discrete frames - for motion, this would be a temporal blurring filter - not sure how you acheive that!
Cheers,
David.
P.S. Can someone split this thread at the point it changed from lossless CDs to frames per second?
http://www.mikhailtech.com/articles/editorials/fps/ (http://www.mikhailtech.com/articles/editorials/fps/)
Do you think that someone really ought to develop an ABX program for frame-rates for First Person Shooters?
http://www.mikhailtech.com/articles/editorials/fps/ (http://www.mikhailtech.com/articles/editorials/fps/)
Do you think that someone really ought to develop an ABX program for frame-rates for First Person Shooters?
I don't think you'd even need a program, just the help of a friend.
-1- Specify anywhere from two to perhaps four framerates per test cycle.
-2- Tester leaves the room.
-3- The friend sets one of the framerates specified in step -1- (may occasionally be the same framerate...i.e., no change).
-4- Tester re-enters the room and plays the game (obviously without looking at framerate settings), and says which framerate he/she
thinks is set. Or perhaps just whether there has been a framerate
change.
-5- Repeat steps -2- through -4- until you get a definitive set of test results.
-6- To try a different set of framerates, goto step -1- and repeat procedure with a different set of framerates in the test cycle.
I'd guess within an hour that the tester would have a good sense of how many (and which) framerate differences they
could detect and how many (and which ones) they
couldn't.
[span style='font-size:7pt;line-height:100%']
Edit: Clarification...[/span]
..... its like frame-rates in video games: once you're over about 60 frames per second, any more is a waste.
....
Wo ho ho, speaking of bs. You think the average person cannot tell the difference between 60 and 100 fps? Do you do much computer gaming?
you think you can? wouldn't it be up to you to prove your claim?
I don't think you'd even need a program, just the help of a friend.
-1- Specify anywhere from two to perhaps four framerates per test cycle.
-2- Tester leaves the room.
-3- The friend sets one of the framerates specified in step -1- (may occasionally be the same framerate...i.e., no change).
-4- Tester re-enters the room and plays the game (obviously without looking at framerate settings), and says which framerate he/she thinks is set. Or perhaps just whether there has been a framerate change.
-5- Repeat steps -2- through -4- until you get a definitive set of test results.
-6- To try a different set of framerates, goto step -1- and repeat procedure with a different set of framerates in the test cycle.
I'd guess within an hour that the tester would have a good sense of how many (and which) framerate differences they could detect and how many (and which ones) they couldn't.
That is a perfect description of a single blind test, not an ABX test. A Double blind would be a better test, though it's difficult to implement in practice.
The key to an ABX test is that you can directly compare X (the test) to A (the control) and B (the experiment)
In any case, I remember an article/post claiming that the frame rate also affected the physics of a virtual world. It was never clear to me how exactly that was possible, particularly in a multi-player battle.
From my (pathetically extensive) gaming experience, the minimum absolute frame rate is the important number in terms of playability, not the average.
It just doesn't matter if you average 200FPS over a timedemo if it drops to 10 in the middle of a huge battle in the center of a large and complex environment. Then it will be a slide show.
Likewise, a 50FPS average frame rate with a minimum of 35 will be very playable.
I don't know... if there was a good, realistic motion blur effect, I doubt you'd need much more than 30 to be transparent. But without motion blur, you approximate a perfect motion blur the faster it goes. I don't know how perceptible that is, but it could be, perhaps more so with training.
That is a perfect description of a single blind test, not an ABX test. A Double blind would be a better test, though it's difficult to implement in practice.
The key to an ABX test is that you can directly compare X (the test) to A (the control) and B (the experiment)
In any case, I remember an article/post claiming that the frame rate also affected the physics of a virtual world. It was never clear to me how exactly that was possible, particularly in a multi-player battle.
Good point, but IMO the process as-is would still accomplish it's purpose: To debunk myths, or at least just to determine an individual's threshold of perceptibility (which would vary between people as does audible transparency).
Another (complicating) factor is that, based on an article I read a few years ago, a person's capabilities of vision vary with level of duress. In a panic, we effectively "see more" than we normally do. Perhaps it's advanced and voluntary (because we
want to see more), or primal and involuntary (because we
have to see more to stay alive), but either way the threshold of perceptibility would vary accordingly with even the same person tested under different conditions.
As people describe things "going into slow motion" in a moment of panic, perhaps that translates into a person being able to differentiate 100 fps from anything slower only moments after they
couldn't differentiate 60 fps.
And, finally, consider "false panic" as well...you're playing Half Life, and in the game you're ambushed. Would you all of the sudden be able to see at a faster rate?
It only happens in real life under artificial lighting, which gives a strobed effect. Otherwise, the wheels only look like they're going backwards when they really are!
I've seen the "wheels going backwards" effect in normal lighting conditions (daytime sunlight). Do I just have really messed up eyes?
I've seen the "wheels going backwards" effect in normal lighting conditions (daytime sunlight). Do I just have really messed up eyes?
Sometimes this can be caused because it's sunny and the hubcaps are shiny. The car is moving by and the tires are spinning, so the changes in angles of light reflection off the wheels can cause various effects. I don't know whether or not it can be caused by our perception of motion also.
As far as FPS goes, there is absolutely no doubt in my mind that the difference between 30fps and 60fps is both noticeable and substantial. I'd wager that the maximum point at which you can notice a difference is something a fair amount higher than 60fps.
Reguardless, you're limited by the vertical refresh rate of your monitor. Typically, that's somewhere between 60 and 120Hz (70-90 being more common). You can't see more than that because your monitor never displays more than that, no matter how faster the computer is rendering. Now, I know for sure I can see the difference between around 100 and 120Hz refresh rate on a CRT (actually did a blind test with a friend once - the difference in the stability of the image was quite obvious). But that doesn't necessarily translate at all to frame rate perception.
Also, you're really limited to some even division of the refresh rate in terms of FPS. If you're monitor is set to 84Hz, you can have a game going at 84fps, 42fps, 28fps, 21fps, etc. It's either that or you've got nasty shearing (the frame is displayed while half of the next frame has been rendered - the "sync to vblank" option forces that not to occur), or some sort of combination of those rates which perceptually won't look much better than the lowest one (unless the lower rates occur infrequently). So if that' the refresh rate of my monitor, I want my game to be going at that FPS, so it's not dropping down to 42 which would be mildly annoying, or 28 which would be really annoying.
It just doesn't matter if you average 200FPS over a timedemo if it drops to 10 in the middle of a huge battle in the center of a large and complex environment. Then it will be a slide show.
Very important point. It is theoretically possible to be rendering thousands of frames per second (your computer can count them, even though your monitor can't display them) when you're in a simple room by yourself. Counting the average doesn't make any sense.
I don't know... if there was a good, realistic motion blur effect, I doubt you'd need much more than 30 to be transparent. But without motion blur, you approximate a perfect motion blur the faster it goes. I don't know how perceptible that is, but it could be, perhaps more so with training.
I'm not in the USA, but there's an American example to illustrate this...
IIRC Friends is shot at 30 frames per second on film (or progressive video) so there are only 30 different images per second.
Normal video (e.g. a live sporting event) is shot at 60 fields per second, and each of those fields (1 frame = 2 fields on TV - look up "interlace") represents a different moment in time - 60 different images per second.
So, Friends is 30 images per second, Baseball is 60. It's not a fair comparison because of interlacing, and obviously there's more fast movement in baseball, but look at any movement closely in friends - it's not quite as smooth is it?
(This example won't work in Europe - Friends is converted to our 50Hz system and looks terrible).
FWIW it's even possible to see the difference on live video footage between the 50 fields per second in Europe, and 60 fields per second in the USA. 60 gives slightly smoother movement (and gives much less flicker, but people in 50Hz countries rarely notice this).
Cheers,
David.
I've seen the "wheels going backwards" effect in normal lighting conditions (daytime sunlight). Do I just have really messed up eyes?
No, I can confirm that I have seen it too.
I can give a more detailed description of the effect:
When a wheel (with the appropriate markings for me to observe the effect) begins to spin, I easily see it as spinning in its intended direction. After it passes a certain limit (not sure if this limit varies from human to human; probably does), I know that the wheel is still spinning in the intended direction, however, I also observe that a holographic-like image (can't think of words that fit closer to my intended meaning) beings to spin backwards, slowly at first, then faster and faster till it also becomes a blur along with the physical revolution of the wheel. This happens multiple times as the wheel spins faster, and the periods between each effect gets smaller.
It's an interesting phenomenon, and I guess that this effect can be calculated precisely given the size, spacing of suitable markings, angular acceleration of the wheel, diameter, etc.
Another well known phenomena: the greater the display brightness, the greater the perceivability of its flicker.
A FPS ABX program would be great, but it would have work in real world applications.
http://www.viperlair.com/articles/editorials/misc/fps/ (http://www.viperlair.com/articles/editorials/misc/fps/)
From my (pathetically extensive) gaming experience, the minimum absolute frame rate is the important number in terms of playability, not the average.
It just doesn't matter if you average 200FPS over a timedemo if it drops to 10 in the middle of a huge battle in the center of a large and complex environment. Then it will be a slide show.
Likewise, a 50FPS average frame rate with a minimum of 35 will be very playable.
Do you know of any 3d games that have benchmarks that will output this value? Is there a way for one to use Half-Life or Counterstrike to run a benchmark and output this value?
Do you know of any 3d games that have benchmarks that will output this value?
The Serious Sam games can output the FPS over time in timedemos I believe.
In the Unreal Tournament games, the stat fps command will give you a real-time output of the frame rate.
Since I've got no motion blur in Unreal Tournament 2003, I can see the difference between any fps as long as I move the mouse fast enough for a bright spot in the game to leave a discontinuous mark on the screen.
What games have motion blur ? Is it effective ? It's obviously the first requirement for a blind FPS test.
Descent 3 has motion blur if you have a P-III or above (I think it uses SSE).
The other day my uncle said to me "why do you set your monitor refresh rate so high? the tv is 60hz". I didnt even bother, I just said "i like it this way." That is what I'm going to do in this thread. I like having >100fps in games. I will say this, however: Most of the people I've talked to about this who claim that anything ove X fps is pointless have actually either had very little experience over X fps, or have never experienced X fps it at all.
...Sort of reminds me of the people who say a 128CBR mp3 is good enough and anything more is overkill.
Do you know of any 3d games that have benchmarks that will output this value?
The Serious Sam games can output the FPS over time in timedemos I believe.
In the Unreal Tournament games, the stat fps command will give you a real-time output of the frame rate.
hmm, really what I meant was can any output the lowest frame rate met, not the average frame rate per second.
I don't know... if there was a good, realistic motion blur effect, I doubt you'd need much more than 30 to be transparent. But without motion blur, you approximate a perfect motion blur the faster it goes. I don't know how perceptible that is, but it could be, perhaps more so with training.
GTA3 and Vice City have this, simply because they were designed for ps2, and run like shite on most pcs (althoug Vice City is a REAL improvement over GTA3 on pc). They are locked at 30fps on ps2, and it is a good idea to do that for pc. I find the motion blur severely limits fps.
Do you know of any 3d games that have benchmarks that will output this value?
The Serious Sam games can output the FPS over time in timedemos I believe.
In the Unreal Tournament games, the stat fps command will give you a real-time output of the frame rate.
hmm, really what I meant was can any output the lowest frame rate met, not the average frame rate per second.
fraps
...Sort of reminds me of the people who say a 128CBR mp3 is good enough and anything more is overkill.
umm thanks dude...i resent that.
i might have said you "sort of remind me" of people that have made various clames about their hearing ability with respect to certain audio compression formats vs. CD, but i didn't.
but perhaps you are right...i retract my statement about anything over 60fps being overkill.
see you on the flip side.
Think about this for a second. Electricity cycles at 50 or 60Hz depending on where you live. Lighting, indoor or outdoor, isn't typically filtered in any way as far as I know. And yet you don't see pulses in the lights.
I'd wager that in a relaxed state, nothing higher than 60Hz or so is required for a very natural feel to gaming or video - however, many games can incur fight-or-flight reactions sometimes, and for this reason, I think gaming sometimes requires maybe 90fps.
Actually some games set the monitor to 60hz, i was referring to a desktop environment in my comment about refresh rates. I defy anyone to look at a 60hz monitor for more than 5 minutes without some discomfort.,
...Sort of reminds me of the people who say a 128CBR mp3 is good enough and anything more is overkill.
umm thanks dude...i resent that.
i might have said you "sort of remind me" of people that have made various clames about their hearing ability with respect to certain audio compression formats vs. CD, but i didn't.
but perhaps you are right...i retract my statement about anything over 60fps being overkill.
see you on the flip side.
I dont mean to insult you. I certainly don't make any extraordinary claims about my senses. To the contrary, I have very acute hearing BUT am a terrible judege of audio quality/artifacts ect.... I use mpc because it encodes the fastest on my machine.
In retrospect, I don't really even think that audio compression and video game fps are very good to compare. My comments sometimes tend to incite and i apologize.
60 Hz monitor flicker is easy to tell from, say, 75 Hz flicker, even in a blind test. But I don't know if 60 fps can be told from 75 fps in a 75 Hz monitor.
Incandescent lights don't flicker, because they aren't less bright when crossing zero voltage, because they have too high temperature "inertia". Fluorescent lights are different, and it's not unusual to see flicker in one of these here in europe (50 Hz).
fluorescent flicker can be pretty annoying here in North America, I can't imagine what it might be like in 50Hz areas, yikes, never thought of that.
why is it that you notice the flickering of computer monitors more if you look slightly to the side of it, as opposed to directly on?
why is it that you notice the flickering of computer monitors more if you look slightly to the side of it, as opposed to directly on?
IIRC, I believe that human peripheral vision is motion-sensitive.
I think it was in Jurassic Park where they make mention of a dinosaur that can only see you if you move/run. Our peripheral vision (vision out the side) is like that. It's speculated that we evolved like that so that we could be more aware of dangers that approach us.
Now that we're civilized, we can use it to determine if the refresh frequency of a monitor is above 60Hz or not
Also, human vision off-center is significantly more sensitive to brightness variations (but less sensitive to details). Astronomers use a technique called "averted vision" to see very dim objects through a telescope. In any given aperature, some dim objects are impossible or very difficult to see when looking straight at them, but looking just to the side, they become quite visible. Google for "blinking nebula" for an example.
just a comment about flickering vs motion:
there are 2 different frequencies, one is called 'enough fps to give the impression of motion' and another is 'enough fps to disable flickering'.
(old example: films shot at 24fps are actualy shown at 24x2 or 24x3 to remove the flickering in cinemas.)
and it's not unusual to see flicker in one of these here in europe (50 Hz).
yes, they are usualy in contra-sync so to get 100 Hz, but when things go wrong...
In any case, I remember an article/post claiming that the frame rate also affected the physics of a virtual world. It was never clear to me how exactly that was possible, particularly in a multi-player battle.
with my limited experience with some 3d apps that actually did the collision detection there was usually a separated unit of time - 'something like mimium delta time between calculations', so you would higher that when the objects would look through each other when colliding...., uhm, so in case that value is the same as fps that could happen. <offtopic> i have also seen a game when you get killed only in replay, but that counts as real end... </offtopic>
fraps
FRAPS is not exactly the highest accuracy, being a third party proggie and so forth. It produces quite repeatable results anyway, IIRC. I believe aceshardware uses FRAPS to test certain newer games.
The Serious Sam timedemo funciton will give you the minimum FPS as well.