Lag Analysis For the PlayStation Move 71
The $64,000 question about Sony's upcoming motion control system, the PlayStation Move, is how responsive it will be compared to traditional console controllers and its counterparts from Nintendo and Microsoft. Eurogamer slowed down videos of Sony's tech demo software to establish a rough baseline latency that developers will have to work with. Quoting:
"While exact latency measurements aren't possible in these conditions, a ballpark idea of the level of response isn't a problem at all. The methodology is remarkably straightforward. Keep your hand as steady as possible, then make fast motions with the controller. Count the frames between your hand moving, and the motion being carried out on-screen. Equally illuminating is to stop your movement suddenly, then count the frames necessary for your on-screen counterpart to catch up. While not 100 per cent accurate, repeat the process enough times and the frame difference becomes fairly evident. Bearing all of that in mind, and recognizing that we don't know how much latency the display itself is adding, I'd say that a ballpark figure of around 133ms of controller lag (give or take a frame) seems reasonable, certainly not the ultra-fast crispness of response we see from games like Burnout Paradise or Modern Warfare, but fine for most of the applications you would want from such a controller."
Not for action games then? (Score:2, Insightful)
Re: (Score:1, Informative)
It's been done for Natal, though maybe not in as much detail. Quick Google search says between 80ms and 120ms.
http://dualshockers.com/2010/02/23/natal-input-lag-timed/
Re: (Score:1)
Not for action games then?
Honest question - what kind of games are motion controls good for besides "action games"?
Bowling, fitness, golf, various other sports games? Seriously, take a look at the Wii lineup and you'll see plenty of games that could be considered action games in the widest sense but usually aren't.
Re: (Score:2)
Not only that, but compounded with network lag which can easily add 50-100ms on top of that. So now you are two frames behind. At 60fps that is fast, but to the twitch fps gamers, it would probably certainly be noticeable.
Re: (Score:3, Insightful)
Since I suppose that 133ms + delay from your screen is quite noticeable.
The 133ms include the delay from the display, so its not that bad.
Too bad if you're sword fighting with a friend, 133 ms can make the difference between living or having your head roll on the floor!
When both are using Move, both have to deal with the lag, so there is no difference.
Re: (Score:2)
There is no difference if you are both reacting to something external, but if you are reacting to your opponents actions, then yes it most definitely will be a difference because your reaction is delayed from when you can perceive the signal.
Re: (Score:2)
The display should only account for 16-30ms (30 to 60fps) of delay. They're recording 133ms or four frames at 30fps of delay, or eight frames at 60fps.
Re: (Score:1)
Sheesh. Some friends YOU have!
Re: (Score:2)
Yay 133ms (Score:5, Funny)
It'll be like playing Quake on dialup again, oh the nostalgia!
Re: (Score:1)
Agreed, 133ms is when you start to swear a bit when it happens bouncing a signal to an online server a few countries away and back. Not from the controller to a tv and then back to the eyeballs.
Re:Yay 133ms (Score:5, Insightful)
Don't forget that games have inbuilt lag compensation, so it doesn't feel like 133ms to the person playing; your gun makes the firing noise and animation straight after you click, regardless of your ping.
Input lag is by far worse than network lag for games.
Re: (Score:2)
, its not huge, but enough to notice when playing fast paced games
You need to have your head examined.
At 60 FPS, a frame is 16.7 ms long. I.E. adding 7ms to your input latency doesn't even show up on the screen.
Re: (Score:2)
At 60 FPS, a frame is 16.7 ms long. I.E. adding 7ms to your input latency doesn't even show up on the screen.
It does if your baseline input latency is 10ms.
Re: (Score:2)
Only if you assume the input is happening at the exact start of a frame (so isn't sampled until 16ms later in both cases). If you assume the controller can be pressed at any point during a frame, then the difference in input latency does show up.
Assuming the game samples input at the start of a frame and the results of that sample show up on-screen 16ms later at the end of the same frame*, the latency between pressing a button on a 1ms controller and seeing the result on-screen would be 17ms if you pressed
Re: (Score:2)
Two frames, plus just under the full input latency. 32.99ms! Pressing it later and later gets you slowly back towards pressing it 1ms before the input sample for a frame is taken.
I like your analysis of the issue, but quite frankly you're really proving my point.
The "Ack, wireless = delay = fuck wireless" was a good argument back when it was noticeable on all wireless mice say 6 or 8 years ago. Now, however, assuming your wireless mouse doesn't blow a boatload of goats, you're talking a maximum visible difference between a wireless and wired mouse being, on average, zero to one frames.
I don't really care how much of a pro/gosu/azn/twitch gamer you are, if you're bitching about on
Re: (Score:2)
I wasn't arguing that 8ms wasn't insignificant, just that it would still exist on top of the frame time. Compared to the 100ms latency of a typical console game, 8ms is minor.
Strangely I haven't seen anyone complain about the latency of a wireless gamepad, only mice.
I personally use a wired mouse, but only because on my wireless mouse the batteries always went flat in the middle of a game. With a rechargeable mouse that wouldn't be an issue.
Re: (Score:2)
on my wireless mouse the batteries always went flat in the middle of a game.
Logitech makes a couple of mice with removable batteries that charge in the mouse dock. Good for solving that issue.
Personally, I use a bluetooth mouse, but I get around the dying battery issue by having two of them. MX-5000 desktop was on clearance for $30 at best buy some years back, so I got 5 for the price of 1 and outfitted two laptops, desktop, HTPC, and so on.
Re: (Score:2)
I have seen captured video of pro quake and counterstrike players in which they reacted to something, aimed, fired, and aimed back on their original course, all within the span of a single frame. That is, the crosshair was never rendered on-screen as pointing at their target.
Re: (Score:2)
I have seen captured video of pro quake and counterstrike players in which they reacted to something, aimed, fired, and aimed back on their original course, all within the span of a single frame. That is, the crosshair was never rendered on-screen as pointing at their target.
Bull shit. Sorry, you might have thought you saw that, but it didn't happen.
Re: (Score:2)
A) False. Consider a simple game in which you press the right arrow when you see a light flash, and your character begins moving to the right at 10 pixels per frame. When you see the light flash, assume you have zero reaction time and press the right arrow immediately. With no input latency, the next frame will show your character moved 10 pixels to the right. With half a frame of input latency he will be moved only 5 pixels to the right.
B) You also assume here that reactions are only made to visual sti
Re: (Score:3, Interesting)
So we're clear, we're talking about input lag here.
"lag compensation" isn't what you're describing, by the way. That's just lag hiding. Lag compensation would be having the server wind the world state back (approximately) to where it was when the client performed an action, in order to determine the result. Do you know of any games that do that?
Re:Yay 133ms (Score:4, Insightful)
Someone posted a humor article describing how JFK was actually around a corner at the moment his assassin fired, but I couldn't find the link.
Re: (Score:3, Informative)
Re: (Score:1, Interesting)
"lag compensation" isn't what you're describing, by the way. That's just lag hiding. Lag compensation would be having the server wind the world state back (approximately) to where it was when the client performed an action, in order to determine the result.
Do you know of any games that do that?
Every popular FPS made in the last 8 years. Adjusting target lead by latency is an ancient memory. Tribes2 even did it, back in 2001.
Re: (Score:1)
However unpopular FPSes like Section 8 and Global Agenda (notably both involving jetpacks and thus highly mobile targets at times) don't for some reason.
Re: (Score:2, Insightful)
That's not lag compensation, it's simply doing some of the game actions immediately on the client side without any host side validation. Without it, basic actions like moving, firing, jumping, etc would all have to be lead, and the amount of lead would vary from game to game depending on your latency to the host (extremely frustrating).
However, these things are only happening immediately on your screen, not the host's console. So although you might see your bullets connecting with the target's face, it n
Re: (Score:2)
The claim that "input lag is by far worse than network lag" is absurd unless you're talking about a game with client side hit detection (e.g. Shadowrun), but most console games use host side detection (Halo, Gears of War, Call of Duty, etc).
The amount of disconnection you feel with input lag is by far worse than network lag, having your mouse take 100ms to turn your view would be unplayable for me; yet I will happily play with 100ms of network lag.
Re: (Score:1)
I would agree if the scenario you speak off was relevant. Respectable developers just don't release games with crippling input lag, which is why I can't see them using the motion driven controller for an FPS until they get the lag down. I'm not so sure 100ms is crippling for most FPS (you can probably play a tactical shooter or medium-paced shooter such as Halo), but it probably would be for twitch shooters like Quake.
My objection was to your phrasing - you made it sound like input lag was the primary fa
Re: (Score:2)
It's when your red dot sight on Quakeworld TF is always a couple of centimetres behind where you're actually pointing on screen.
It's where you have to aim where you think the other player will be in 133ms and can tell yourself you're an awesome player for one shotting him because he was on ISDN and only had 60ms ping and should've easily got you first.
But then you get broadband, and realise it is actually more fun being able to just simply aim and shoot, as much as there was a feeling of victory to be had f
Re: (Score:2)
Definitely a disadvantage, but with games like teamfortress, being able to kill the enemy isn't so important.
Getting the flag from one point to another could be way more important, and the flag moves about the same whether you have 10ms ping or 1000ms ping
And with custom team fortress, coming up with a suitable setup can help a lot to mitigate lag
Re: (Score:2)
Team fortress had pretty good lag compensation as I remember (to the point where people who were not lagging would complain that they died after they had got into cover).
Resident Evil 5 (Score:2)
Re: (Score:2)
Yeah, or like playing WoW on a Canadian ISP using DPI... I'm lookin at you Rogers. -.-
And again my lament... (Score:3, Insightful)
Wii-style controllers are for fun. Not for accuracy. Not for performance. Not for precision. They are meant to be a fun tool to play a game. It is by far the most inaccurate way to put input into your machine. It's like typing on a "virtual keyboard" on a touchscreen, or a projected one. Yes, it's insanely cool to use a keyboard projected onto your desk instead of an ordinary one. But you neither get the tactile feedback, touch-typing is pretty much impossible and so on. You will not write as fast on any projected/displayed keyboard as you do on a standard keyboard.
The same applies to these motion controllers. Yes, they're a freaky cool toy. They sure offer a new aspect of "fun" to games. Especially in company when you can watch others gyrate, bounce and stretch (funny or sexy, depending on gender, body type and personal preference). But be honest: Ever tried to play DDR on a keyboard instead of the dance mat? Dunno about you, but I'm HEAPS better using a keyboard.
In short, if you want performance and precision, stay with the old style controllers. Motion controllers are cool if the way is your goal, i.e. when playing the game is where the fun is at. If reaching the goal is paramount, i.e. beating some boss or making the next achivement, they are rather unsuitable.
Re: (Score:1)
Re: (Score:3, Insightful)
Wii-style controllers are for fun. Not for accuracy. Not for performance. Not for precision.
Is that by necessity or is that just the way it's been so far? Is it impossible to improve on or change the expectations of?
Re: (Score:1)
Input lag is a twofold issue. the wiimote simply doesn't have a very fast connection. 100 frames a second. Information just doesn't get to the system as fast as a button press on a 360 controller.
Input lag is more than just how fast an accelerometer reading gets to the system. Games that require complicated gestures simply can't determine if you have done the gesture, unti
Re: (Score:2)
Well, now I really do grasp how quaint the Wiimote is...
Re: (Score:1)
Wii-style controllers are for fun. Not for accuracy. Not for performance. Not for precision. They are meant to be a fun tool to play a game...
Pardon my ignorance, but I thought that games were suppose to be fun, or maybe I'm just not a hardcore gamer.
Re: (Score:1)
Wii-style controllers are for fun. Not for accuracy. Not for performance. Not for precision.
"Fun" and "control" aren't mutually-exclusive. Accuracy and precision are characteristics of a specific control sensor, not a mode of control. You hold thumbsticks in such high esteem, but the joysticks of yore required frequent recalibration, and had issues with max ranges and zero deadzones. As the technology matures (e.g. WiiPlus), so will the precision.
Motion controllers are not very good for binary input; that's what buttons are for, and that's why the Wii has gotten so much flak for "waggle" games. Co
Re: (Score:1)
More importantly sometimes precision is not desired. Look at all the crazy abstractions golf and bowling games came up with before the Wii in order to simulate the difficulty of getting an accurate shot done. We see nonsensical results when games try to pass a difficult action (like certain sword moves) off as trivial and just demand gesture activation under the assumption that you can do that at any time. I think to make full use of motion controls we have to ditch many of our current gaming conventions be
Re: (Score:2)
Why is it that fun is the antithesis of accuracy and precision? Accuracy and precision are requirements for fun.
It would be no fun to play Monopoly if I rolled a 6 but the game moved me 5 spaces.
A shooting game would be frustrating if I shot the target but it says I miss.
A fighting game is no fun if I make a really good move in the nick of time, but my character loses anyway.
Re: (Score:2)
Generally by the time developers get their hands on something, it's set. You're not going to see any improvements in that response time in this gen of tech. What's worse is the 133 is going to be on top of any network latency, meaning this is going to be well-nigh impossible to use for any multiplayer game that isn't split-screen. Most twitch-based shooters are balanced at 60-100 latency since that's the average connection. 200 and you start to notice shit. This will be *starting you out* at 200, the 60 fro
Well, okay then. (Score:2)
I guess we can compare these results to when the product is actually released for more than piddling around.
50 ms max (Score:2)
Having gone through a series of monitors with different input lags and tested my own tolerance, I can't deal with a total of more than 50 ms input lag + rendering before I'm bothered and 75 ms before I lose my mind and spend my time on slashdot instead.
Lag, jitter and tracking (Score:2)
Also, just because the camera is pointing at the user doesn't mean it's capturing their movements at 60fps. I imagine the sample rate is tunable depending on the kind of game being played. Some games may only "sync" with the ball every 10th of a second, others more, others less and som
Re: (Score:2)
I guess all these things are tweakable and it's up to games to make the right trade off between accuracy and responsiveness depending on what kind of game they are.
You *guess* they're tweakable. Okay, let's go with that premise.
Name a game where you'd be willing to have an input lag of a half-second for perfect accuracy. Pretty much any game-type you can name would be better played with either a controller or a mouse.
Now, name a game where you'd trade accuracy for instant responsiveness. Again, any game that requires instant responsiveness is going to also require accuracy to a high degree, and would be better served with either a controller or a mouse.
For the real ga
Re: (Score:2)
First off the latency wasn't half a second, it was estimated at 133ms even with unknown factors such as the latency of the TV, game settings and unoptimized code to consider. Second, plenty of casual or sedate games would benefit from motion sensing that tried to smooth out spikes caused by wobbly hands or whatever.
133ms is huge (Score:2)
I spent $700 on an IPS panel monitor a few years ago and am an avid gamer. Anantech, after constant complaints about input lag from many readers, began testing it in their monitor reviews. My monitor scored around a 50ms input lag.
That 50ms forced me to buy another monitor. My clicks lost accuracy in RTS games, my aim was off in FPS games, and I could see the delay when watching movies (lip movement didn't match voices). I bought a TN panel monitor of the same size and have them next to each other now,
Re: (Score:2)
I gave up on LCDs altogether after discovering that the decent ones all have either bad input lag, horrible black levels, terrible image quality at non-native resolutions, or all of the above. I'm using a 9-year old CRT because I can't find a monitor that doesn't make me want to throw it at a wall.
Input lag is a serious problem that often goes unrecognized. I played a collection of old Sega games using a wireless controller on a friend's PS3 hooked up to a large LCD TV and I couldn't believe it. Sonic the H
Re: (Score:1)
When running at full speed, my reaction time + input lag was greater than the amount of time it took a spike or robot to cross the screen and hit me
For me that's true even with zero lag.