Yes, but how many FPS do we actually see? The Nth iteration of an always fresh debate

mikiex said:
Douglas Trumbull when developing Showscan found that 72fps was the the peak before there was not difference to the viewer. Although for technical reasons again they used 60fps.
Click to expand...
When you say found, was that experimentation or calculation? ToTTenTranz was suggesting the ideal framerate was determined logically rather than experiementation, which I don't believe. I'd also argue that the idealframerate depends on content being shown including speed. A camera filming a fast pan at 72 Hz may look smooth due to natural motion blur, whereas a computer game could look a bit juddery. If working with perfect frames and no motion blur, the required framerate to appear smooth will need to be faster. I've never done any experiements trying different refresh rates or the like, so I have no personal insights to contribute. ;)
 
I thought everyone did the frequency experience in physics course.
You get a stromboscope (switching light on/off at given frequency) and increase the frequency until you don't notice the dark anymore, trust me it's way above 60Hz (more like 100-120Hz), younger are more sensitive than older too.
(which is usual for any input :p)

I got a 120Hz "LCD" screen it makes a different, but I cannot quantify it.
 
jlippo said:
Would it really be that bad with something like ~16ms delay and 8ms refresh?
Click to expand...
Add in vsync and tripplebuffering and the delay gets pretty severe and quite definitely annoying when the blur is slightly behind your eye movements.
 
Rodéric said:
I thought everyone did the frequency experience in physics course.
Click to expand...
Never did that, but even then that's only a single reference subject - a solid white/black. A large dot quickly moving across the screen, or a camera pan of mountains contrasting against a bright sky, with discrete images and space in between as they move, will show what's necessary to achieve smooth motion. Considering a subject travelling 1000 pixels horizontally in 500 ms on a 1080p TV travelling across so many degrees of the viewer's FOV, how many frames are needed that it appears as a smoothly moving and not jumping across or leaving ghosts?

Perhaps the real answer isn't so much frames per second, but samples per distance? An object travelling a distance of n pixels per second on a screen will need something like n or n/2 samples per second, depending on pixel size.
 
Shifty Geezer said:
Perhaps the real answer isn't so much frames per second, but samples per distance? An object travelling a distance of n pixels per second on a screen will need something like n or n/2 samples per second, depending on pixel size.
Click to expand...

hm... not 2n samples per second? ("SHz" Shirts :eek:)
 
Shifty Geezer said:
When you say found, was that experimentation or calculation? ToTTenTranz was suggesting the ideal framerate was determined logically rather than experiementation, which I don't believe. I'd also argue that the idealframerate depends on content being shown including speed. A camera filming a fast pan at 72 Hz may look smooth due to natural motion blur, whereas a computer game could look a bit juddery. If working with perfect frames and no motion blur, the required framerate to appear smooth will need to be faster. I've never done any experiements trying different refresh rates or the like, so I have no personal insights to contribute. ;)
Click to expand...

Based on people hooked up to equipment to monitor response looking at a screen. At the time cost forced it to become consigned to themepark rides, now framerate is probably the next step in elimnating motion artifacts in films - especially 3d. Thats not to say artistically every film should be shot at 60fps.
 
It's not only fps, it's also (input) latency and refresh rate.
I'm not sure you can analyse one w/o the others on a computer screen, maybe with a projector.

(I was more thinking about screen frequency in my previous post.)
 
AlStrong said:
hm... not 2n samples per second? ("SHz" Shirts :eek:)
Click to expand...
No, I'm thinking two 'pixels' movement per frame is enough to look smooth. Could probably even get away with more than that. Where a pixel is actually a measure of movement across the retina, so of course it changes with screen size and resolution and viewing distance.

Although conversely, at slow speeds, without sufficient resolution than movement can look jerky at 1 pixel per frame, and you really need sub-pixel movements.
 
Forced vsync sucks. ruins the gaming feel. I don't care I can't tell the difference visually between 60fps and 90fps but I can feel the difference in response to control input.

fps caps introduce latency. don't do that, gamers don't like it.
 
FPS cap, like most things, should be optional. I prefer a steady 30fps over wild fluctuations, but others might not. We can all have it our own way.
 
What about Touchscreens.

if you were pulling bits of virtual paper around with your fingers on a microsoft surface what fps would you need for it to be completely convincing.

motion blur isn't desirable there because your eyes might be tracking your fingers or the background

having said that, has anyone ever suggested tracking eyeballs to figure out which piece of the screen you're tracking ... stream textures when you blink :)
 
You must log in or register to reply here.
Back
Top