The "die one second later around the corner" bug
Jones108
Join Date: 2012-12-10 Member: 174670Members
Join Date: 2012-12-10 Member: 174670Members
I dont even know how to call this one. You are an alien running around the corner, one second later, you die. I think everybody knows the situation and im not very educated about the whole topic myself. All i know from experience is, that it never tend to happen that often in other games. Our pings are around 50 or below it.
Comments
If you die around a corner, you died .5 seconds before but just doesn't realized it.
If a skulk is running at you, strafe .5 seconds before he is reaching you.
The attackers reality is preferred. This way you always hit at what you are aiming. Without lag compensation you would need to aim ahead of your target at a ping-dependent distance. This would be even less fun.
You just need to understand, that connections over the Internet can't be instantaneous. It's just physically not possible. So you need to bare with illusion-technics like lag compensation, that try to convince you, the ingame-world would be at the same time for everyone.
heh heehe, false. Remember all those death you've had after crouching behind a cover and still die a second later....? he heeh he *cough* BF3 *cough*
Psst, let me tell you a secret... It's not a bug. But don't tell anyone.. ssssh :-X
Well somebody told me it happened alot in the alpha, but now they fixed it somehow?
Also i dont play BF3. Never gonna touch anything from Origin/EA.
So change the setting. I wanna see what happens.
When that ping-dependent distance is 50ms or less it would definitely be "more fun" than being 500ms behind in dodging shit.
When players can move faster compared to most other games, it makes it seem like the "getting killed behind a wall" is more prevalent, because the speed of your character excaberates the effects of lag compensation.
People seem to forget these 2 important things when talking about lag compensation.
Supposedly, they based their net code on Source.
I can bhop faster in cs than wall spam cele skulk and still die in terms of time much much faster after I'm fully covered around a corner. I'm pretty sure the compensation or whatever is much higher in ns2 than in source etc games.
Meeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeehhhhhh maybe
The bhopping thing could be a little misleading though. Its more than just speed that would cause this "problem" to manifest. For example, even if you died around a corner each and every single time you bhopped, its probably unlikely that you would perceive the problem because bhopping is not a common occurence in CS. You also tend to die in far fewer shots in CS, making you less likely to perceive death after attempting to be evasive.
The issue here, in my opinion, is entirely perceptive. Due to the gameplay involved, people perceive this problem far more easily. That is to say that the same thing happens in the exact same way in many other games, but its far harder to perceive due to consistency and frequency. For example, in this case people think they are "dying a second later" after being around a wall. If they were indeed dying an entire 1000ms later, such an issue would manifest in many other ways. In reality, people are likely perceiving their distance as time. They are so far around the corner that they perceive it as being a long time since they rounded said corner, when in actual fact its just their speed being high that causes this.
To decrease this discrepancy you'd have to increase network traffic by about double, essentially bringing a server to it's knees as far as performance goes.
First optimize the game and the network traffic, and then you might be able to do this.
For the record, IIRC, Max said interpolation doesn't apply to spark in the same way it does in other games. If you made interpolation only 20 ms instead of 100ms, only AI units like macs and drifters and arcs would benefit, not players.
Rather hard to explain this one but I believe someone allready made a thread about this like a month ago (with a youtube video).
Can it be, that you mix up interpolation and server tick rate? I remember Max saying exactly this over the tick rate. But an interpolation reduction (if the server could hold the higher load), would reduce those effects.
Very much this.
I do not understand how there are so many people just saying HTFU, deal with it and that is it normal
Yes, lag compensation and interp are in the majority of mp games these days, and there will always be a discrepency between what each player sees on the screen, and what the server thinks the current game state is. However, I have never experienced these issues to the same extent as with NS2. This is worse than playing NS1 on a 56k dialup.
Your argument that its a fast paced game and therefore servers cannot keep up is not valid. NS1 issues were not to the same extent and gameplay was a lot faster with bhop etc. TFC, a game which was even faster than NS1 in terms of raw aerial speed you could achieve through conc jumps was handled fine using the Goldsrc engine. If what you are saying is correct then TFC scouts should have been almost impossible to hit simply because of the netcode and not through lack of skill, but this clearly isnt the case.
We do not gain anything by sticking our heads in the sand and saying "oh its to be expected, NS2 is just too fast for the current standards".
It would be nice if UWE opened up their engine code for us to take a look at but I highly doubt that is going to happen.
Increasing the tick rate wouldn't affect players because it's decoupled from move rate. There is however potential for gaining responsiveness by increasing the move rate. Before the big net code overhaul shortly before launch there was no limit to the move rate. Clients would send updates every frame, which obviously tanked the servers as client performance increased and the servers got more updates to process. It has now been set to 30 to match the tick rate.
When we're talking about interpolation, what we're interested in is the update rate. What interpolation does is smooth out the transition between one update to the next on the client. In order to do this the client needs to stay one update behind. I believe the servers send updates 20 times per second currently, which would mean ~50ms. We also need to account for packet loss and the fact that the client and server don't update in sync so we double that to 100ms.
So basically, in order to decrease the interpolation we need to increase the update rate. And in order to do that we need to increase the move rate or the server won't have any new info to send. And this won't happen util server performance is increased.
Totally agree here, NS2 just feels... a bit slow and gooey are the best terms to loosely describe the feeling. The "Shot around a corner" is merely a symptom of something else going on in the code, but this cause will be massively affecting every fight in the game, whether or not you realize it.
The server backtracks the position of Xearo to the time when you shot and then does the hit check. Since players are seeing an afterimage because of interpolation but their shot at this afterimage counts, these "shot around the corner" or "bit from a mile away" situation happen.
You cant do much about interpolation, the alternative would be players teleporting around every 50 milliseconds. That would look horrible. The lag compensation on the other hand does not belong to a game, where movement is so important.
If you want to look up more about lag compensation, http://www.ra.is/unlagged/
Luckily the interpolation delay can and most likely will be lowered in the future, but it's not as trivial as simply changing a number.
Lag compensation is bad in games where movement is important because dodging becomes almost impossible. This is even stated as a limitation of lag compesnation by one of its inventors.
I love you.