The performance of the Spring engine

[Neddie]

I don't believe anyone can see a difference between 40fps and 50fps.

Someone should do a blind test to compare just to see what others see

I can see a difference.

[Vadi]

OK, I know people claim that. I simply don't believe it though!

Some simple math such as 50fps = 1 frame every 20ms, 40fps = 1 frame every 25ms, and 5 ms difference is an _awfully_ short time. It's 200th of a second, programs and wikipedia offers a nice comparison of what that means (http://en.wikipedia.org/wiki/Millisecond).

I simply don't believe a human eye can tell a difference in 5 ms, sorry!

[SwiftSpear]

I can't see the counter when I'm jumping. I'm focusing on the landing. It's usually not till after the fact when I've missed a jump 5-6 times that should be easy that I look down at the counter and see the FPS isn't at 100 solid.

Although, to be fair, it's more than eye ability in this particular scenario. KZ jumping requires fluid character motion and extremely precise coordination between mouse and keyboard input. A lower frame rate means the computer is accepting commands less often as well as displaying less actively. I could probably play comfortably at 30 FPS visually if the game would update it's information at 100 FPS.

I can tell the difference between 30 FPS and 60 FPS in a standard PC game. I used to average 30 FPS in HL2 and Doom3, and the games were barely playable IMO. On my new PC the FPS drifts from 60-100, and the quality of play has jumped up markedly.

Oh, and PC motion blur isn't the same thing as TV motion blur. It's not used as a technique to hide FPS, it's a prettifier. Effectively, it chalks up to fast moving objects having little color tails drawn behind them, and it's usually relative to character position, not screen movement.

[edit] Vadi: The question isn't weather the human eye can accurately see a 5ms difference, the question is weather that difference is noticeable when repeated thousands of times. No one could accurately tell you the framerate when presented some random scene unless the framerates are really low. It's an issue of weather you can see any difference at all.

[Vadi]

That period of time has to be short to make any difference - for example I don't think you'll see one if you watch 40 and then 50 fps for an hour. So lets take the same old one second, which is quite big for these purposes. That's 10 frames of a difference, or 50ms, evenly spread out over the 1000 milliseconds.

Newp, sorry, not convinced

[Ixoran]

That period of time has to be short to make any difference - for example I don't think you'll see one if you watch 40 and then 50 fps for an hour. So lets take the same old one second, which is quite big for these purposes. That's 10 frames of a difference, or 50ms, evenly spread out over the 1000 milliseconds.

Newp, sorry, not convinced

God damnit lern2science.
Or google.
Or Think.

The USAF, in testing their pilots for visual response time, used a simple test to see if the pilots could distinguish small changes in light. In their experiment a picture of an aircraft was flashed on a screen in a dark room at 1/220th of a second. Pilots were consistently able to "see" the afterimage as well as identify the aircraft. This simple and specific situation not only proves the ability to percieve 1 image within 1/220 of a second, but the ability to interpret higher FPS.

If you don't want to read it all, then the short answer is, the human eye / brain combination can see well over 100 frames per second and thus far the limits have not thoroughly been tested yet. Suffice it to say, IT IS NOTHING LIKE THE 24, 30, 60 or even 100 fps crap, that gets spouted on the Internet.

Thanks for helping me find some interesting articles, But next time do your own research.

Oh, and let's not debate physical response time or speed.
Because that with training can be faster than our eye can see. do some research on Bruce Lee or something, He could move faster than most cameras could capture him in his time.

EDIT: The second article is pretty much a re-reference to the first article, But whatever, I think I makes a good point.

EDIT2: A real second reference
http://www.firingsquad.com/features/fac ... efault.asp

[AF]

Lets say the human eye has 1 million receptors. A random figure thats probably absurdly accurate by coincidence or wildly innacurate.

But, this gives a theoretical maximum of 1 million, but even then whats to say one receptor didn't capture an extra frame while the other receptors where all receiving at their individual unique times 1 millionth of a second apart at that precise moment?

Eitherway we do not percieve reality at 1 frame per second and I wouldnt be surprised if it turned out that sometimes large portions of what we can see arent updated at all for seconds at a time and that the image we see is not all updated at the same speed.

The very notion of an eye having a frame rate at all is ludicrous. Just wave your hand infront of a CRT monitor in the dark and youll quickly see that to get that you need a low frame rate for your eyes, yet that doesn't explain how we see faster things without that problem all the time.

And I havent even touched on the whole 'the human eye doesnt see whats infront of it, it sees changes infront of it'

[TradeMark]

at first post: If you gonna test unit rendering performance, at least use more than 20 units, because 20 is so little amount it doesnt even change the FPS.

to the human refresh rate craptalk: i was disappointed when i bought my LCD which can only do max 60FPS, while my old CRT did 100FPS, the difference is almost 2x! and someone was saying i cant see difference :D:DD::Ddd ...because he just had retarded eyes/brain

i feel bad for those people who cant see the difference its much more enjoyable to see 100fps than just 60 or less... maybe they just needs to train their eyes/brain...

[ironized]

:O

i get 20 and im running

a

8600gt 512mb ddr3
2gb ddr2 800mhz
AMD 6000+X2!!!!1

now, i was under the assumption that i would get better fps on medium with this system then with our

2.8p4 amd 9600 pro and 1gb ddr ram.....


i was wrong...


elaborate.

[imbaczek]

GF 8-series don't work well with spring, unfortunately. Some people manage to make theirs work, but there's no 100% solution.

[AF]

As lathan stanley noted shortly after geforce 8 series was released, framerates under them are strange.

For example you get low framerates yet upping the water renderer barely makes any difference to the frame rate, he even noted that upping the anti-aliasing from 0x to 16x made very little difference to the frame rate, its totally counter intuitive.

[KingRaptor]

on frames per second:
http://www.100fps.com/how_many_frames_c ... ns_see.htm

[Teutooni]

My 8800 GT works fine for spring... I can get around 20-60 with shadows, 16x AA, and exactly 499 arm hawks circling over dynamic water (about 20 when looking at the hawks and 60 looking elsewhere).

As lathan stanley noted shortly after geforce 8 series was released, framerates under them are strange.

For example you get low framerates yet upping the water renderer barely makes any difference to the frame rate, he even noted that upping the anti-aliasing from 0x to 16x made very little difference to the frame rate, its totally counter intuitive.

I guess that is because spring needs alot of raw computing power and memory due to the huge unit numbers ingame, not fancy shader processors etc. You can up the AA as much as you like, the gpu and memory will be the bottlenecks.

[rattle]

You can only see ~24 frames per second, so a framerate of about 30 is more than enough. I can see a difference between 30 and 60 as some games become a bit smoother, though this depends on the game. Anything above 30 fps is a waste of energy IMO.

[Kloot]

Sure, until you try to do a fast 360-degree turn in Quake.

There's another big reason why the "24FPS ought to be enough
for everybody" myth is quite patentely false. Not only does the
spatiotemporal distance between any two successive rendered
frames decrease as the framerate increases (which provides a
better illusion of continuous motion, since frames are discrete),
but a higher framerate also means that each simulation frame
can take longer to calculate before the impact becomes visually
noticeable. Since in Spring and most other games simrate and
FPS are directly related, a high FPS is a huge benefit even if your
monitor cannot display every frame that's drawn.

[Crayfish]

The USAF, in testing their pilots for visual response time, used a simple test to see if the pilots could distinguish small changes in light. In their experiment a picture of an aircraft was flashed on a screen in a dark room at 1/220th of a second. Pilots were consistently able to "see" the afterimage as well as identify the aircraft. This simple and specific situation not only proves the ability to percieve 1 image within 1/220 of a second, but the ability to interpret higher FPS.

This doesn't really show frames per second.

Darkness is an absence of stimulus. Retinal photoreceptors are activated by photons, or light. This just shows a massive activation of photoreceptors which, in the absence of other stimuli, persists as an afterimage.

This is what allows us to see low fps cinema films as continuous scenes, the darkness inbetween frames (an absence of signal) is overridden by the afterimage of the bright frame before it.

In this air force experiment, the image could have been shown for any length of time (provided that it was bright enough relative to the dark room to activate a sufficient number of photoreceptors). It's the afterimage that allows the pilots to identify the planes.

A real test of visual processing 'fps' would be to show the image between other frames of equal brightness at 220 frames per second, asking the pilots to look out for the plane and identify it. I'd be interested to see the results of such a study. I suspect that the term 'fps' is an oversimplification when applied to human vision and would vary depending on the context.

Hmm, just read this article (posted by KingRaptor) and realised it says the same thing, better.

[Tobi]

Kloot: Actually I'd say in Spring simulation rate and FPS are quite unrelated, except that they compete for the same resources (CPU and memory).

If FPS >= 30 for example, there's no correlation at all between sim rate and FPS anymore, because sim rate always is 30 updates per second. (at speed 1.0).

[Kloot]

The resource contention was what I meant yeah. If each
simframe requires say 25 milliseconds at speed 1.0, the
renderer only has 250ms left to work with and FPS will be
much lower than if the simulation frames all finish in half
that time. If you look at it from the renderer's perspective
then anything over 30 FPS means the 1.0 simrate can be
maintained, but there's still a difference between 30 FPS
and 120 (for instance, when zooming in or out using tab)
and the more FPS you start out with the more time each
simframe can claim before maintaining the simrate really
becomes problematic (since so much time is available for
rendering).

[smoth]

I guess that is because spring needs alot of raw computing power and memory due to the huge unit numbers ingame, not fancy shader processors etc. You can up the AA as much as you like, the gpu and memory will be the bottlenecks.

I have 3 gigs of ram

[Felix the Cat]

A lower frame rate means the computer is accepting commands less often as well as displaying less actively.

Not necessarily, simulation and rendering can be completely linked or nearly completely independent depending on the game's architecture. Most games out now tend toward the latter.

[Vadi]

We're also forgetting the input, which all gamers are claiming is oh so important down to that 5ms. I wonder what's the response rate on typical hands

Btw, that pilot research, imo, does not apply. I'm pretty sure you'll be able to tell a difference in a movie with, lets say, a wheel spinning very rapidly. But not in a guy jumping and running about a human speed in an fps game. The article fails to say if the pilots saw a difference between the image being flashed at 1/220 (that's 220 fps) and, lets say, 1/200 (that's 200 fps). Just because the eye's visuals managed to hit once in the many flashes, and the pilots remembered an image that the eye caught, it doesn't prove that someone will be able to see a difference in a movie at those, or even lower, rates.

Read your own arguments next time, because they fail