Leading for ping is ridiculous, Mk.2: An Example

[dazman76]

With prediction you never get killed while behind cover. This is a fallacy.

No, it isn't Player A is the shooter, Player B is the target. Player B has a good ping of 50 and is close to being in cover. Player A has a ping around 200ms, but it spikes briefly to 500ms right as Player B makes a slight adjustment to their heading, before entering cover. Player A takes the shot and their client is then authoritative - even though Player B has moved just enough to evade the shot, that isn't how Player A's client sees it - so the hit is registered and sent to the server, and Player B dies just inside cover.

This is what is being shown, in the CSS videos linked

 

[M55ikael]

Favoring the shooter is the entire point, though. In RO, we have a game design where the projectile speeds versus character movement makes dodging impossible. By the time you even realize you're under fire, it's already been decided if you're hit or not. If we were playing something in more of a shmup vein like Descent, where weaving through slowly moving fire is a huge part of the gameplay, then yes, it would be problematic to make the shooter authoritative, but in RO, everything is already decided on the shooter's side to begin with, so giving them full fidelity comes at no cost.

As for the larger theory discussion, this is hardly a new topic. Client authority has been a staple of internet gaming since before there was internet gaming. All the earliest online action games on GEnie, Compuserve, AOL, Gamestorm, etc were made possible by authoritative clients. It's only a fairly recent development that computing resources and bandwidth have been inexpensive enough to allow the luxury of fully server-based models for a wide audience. Even now, authoritative clients are still necessary for large volume settings like MMOs, to cut down on server workload, and for high-precision environments like combat flight simulators, because of the prohibitive complexity and overhead of doing server history rollback on a precise level with anything other than hitscan weaponry and the sheer pointlessness of not using any latency compensation at all. RO2 shares a lot more in common with a flight sim than it does with a traditional shooter like UT.

As for if it can be done, it should be fairly trivial in Unreal Engine. Just run the firing logic on the clients, like in single player, and replicate the hit functions over to the server. It might even be possible to drop in as a mutator. (Hmm......)

Now this is what I'm reading the forum for. Not brainless banter and bickering. I'm not saying what you wrote is law, but the important thing is that it's well thought out.

 

[The_Cook]

Yeah, lower pings mitigate the problem, but nothing can cure it entirely. It exists in a small amount even on a LAN connection and its sub-1ms network travel time, thanks to the server tick rate delay. It's just inherent to the design. That's why virtually all multiplayer action games use some sort of latency compensation method.

"Play on a server with a better ping" is, unfortunately, not a viable option with a niche game like RO2. For the last several months, there has been just one server with a consistent population on my entire continent. It's on the Atlantic coast, I'm on the Pacific coast. That's not a recipe for blazing connections.

It's also not the real source of the problem, anyway. Almost every other FPS ever made will eat 150-200ms for breakfast, and most can smoothly handle two, three, or even six times that much latency. The RO series is special in this regard, and not the good kind of special.

I've been playing Multiplayer FPS since the first Quake engine. And never EVER has playing on a server with 100+ms ping ever been acceptable. For any FPS to work well and for you to be able to compete competitively you need at bare minimum 80ms. 30-40ms is your sweet spot sub 30ms you are godlike.

The rare occasions when I can play an RO server with 30-40ping the game works correctly and consistently. Crying that things aren't working right when your ping is +100 is like crying that your car isn't working right cause you don't have air in your tires.

 

[The_Cook]

Since when was zig zagging considered a cheese tactic to take advantage of internet latency?

I stop randomly crossing large open spaces.

 

[Mekhazzio]

For any FPS to work well and for you to be able to compete competitively you need at bare minimum 80ms. 30-40ms is your sweet spot sub 30ms you are godlike.

Sure, because Red Orchestra is such a smash hit in the competitive FPS scene that it makes perfect sense to focus on a networking model aimed solely at that, at the expense of basic playability for everyone else in the world?

That perspective is is flawed anyway, because in RO2, while it's less likely to become a total miss, 30ms error is still enough to turn a kill shot into an arm shot, an outcome alteration that is utterly unacceptable from a competitive gaming POV. The game's precision is such that, even on the very best connections, it still needs latency correction, to deal with time-to-next-tick fluctuation if nothing else. Look at the poster children of competitive PC FPS games: Quake, UT (snort) and CS. Of the three, which does RO most closely resemble? Of the three, which is the one with latency compensation? Exactly.

 

[PhoenixDragon]

This is also why you don't get killed behind cover in RO2 and in other games you do. Lag compensations are like hax for those who can't have a good connection.

See, this is the problem with this discussion. We even have a video in this very thread showing that you can still "die behind cover" in RO2, yet people still think it can't happen. Though it does illustrate how little of a problem it actually is, when people don't even recognize it when it happens.

The only disadvantage people have been able to cite for a client-side model, that of "dying behind cover," already exists in the game. It would not be adding a new problem to the game because it's already there.

Well, there's also the idea that it would be easy to cheat by making a man-in-the-middle hack that turns every shot into a hit. That's a fiction. Though I'm not quite willing to say that has never happened, I've never heard of it being done even on massively-popular games with client-side hit-detection. Aimbots are far simpler. Plus, how paranoid do you have to be? We already have VAC and PunkBuster catching anyone who tries anything funny, we don't need to wreck the shooting for the legitimate players to make it even harder for a tiny number to cheat! Particularly when they won't be able to play after a day or two anyway.

I've been playing Multiplayer FPS since the first Quake engine. And never EVER has playing on a server with 100+ms ping ever been acceptable. For any FPS to work well and for you to be able to compete competitively you need at bare minimum 80ms. 30-40ms is your sweet spot sub 30ms you are godlike.

The rare occasions when I can play an RO server with 30-40ping the game works correctly and consistently. Crying that things aren't working right when your ping is +100 is like crying that your car isn't working right cause you don't have air in your tires.

More like complaining that your car isn't working right because the steering is designed to make it completely unstable at over 15 MPH. If you can get where you want to go at that low of a speed, sure, the car is great. But for most real-world users, it's a horrible design.

RO2 is a niche game. The population of the servers show that clearly. There is one populated server in all of North America, and the past day or two, even that's been a bit sparse. If you aren't lucky enough to be in the same state and on a good connection, you're not going to reach your "absolute minimum" latency. The median on that server is around 150ms. That means that most people in North America are screwed if they want to play the game. It is not a good design decision to simultaneously make a game that will have a niche audience and require that audience to live close to a server in order to have a good experience. Look at the comments on RO2 from other places, such as RPS; the number one complaint is about the network model ruining gameplay for most people. If you want the game to only be playable for people with 30-40ms latency, you've just told the vast bulk of potential customers to not even bother trying, and that's the message that's going around the internet about RO2.

Though your expectations for latency are absurd anyway. If your experience had ended with Quake 1 (Or if since then you've only played Quake or UT-engined games, with no latency compensation), then what you describe might be acceptable, but most games are not that dependent on latency. The vast majority of FPS games, ones that use client-side hit-detection or server-side latency-compensation, are perfectly playable at well over the pings you're describing. If you're at 250ms latency, you might be at a disadvantage to someone with 50ms latency in, say, TF2, but you can still land shots just fine. The gameplay still works. Same for Brink, or Nuclear Dawn, or the Left 4 Dead series, or the Serious Sam series. ALL games with much faster movement, and all with a network model good enough that even when the roommates start up a torrent and kick your latency up to 900ms, you still have a fair chance to hit what you aim at - and no matter what your latency is, you aim at the exact same point.

 

[Graphic]

They need to make hit reg client side like *gasp* Call of Duty. I'd rather die after I run behind a wall sometimes and feel like I'm in complete control of my bullets than be able to dodge bullets like a Matrix agent.

 

[Golf33]

The CS video shows why lag compensation is a good idea.

Without compensation, the guy with low ping who is popping in and out of cover to take his shots is effectively invulnerable to the high-ping guy who is watching from a covered, elevated position.

The low-ping guy lambs out and takes a shot. Because he has a low ping, his shot will go somewhere near where he aims (especially since his target is basically still).

The high-ping guy knows exactly where the low-ping guy is and has his sights laid right on where his head will appear. When it does, he fires but by the time he has seen it (thanks to his higher ping) the low-ping player has already taken his shot and is snapping back into cover. By the time the high-ping player can react and fire, according to the server his target is no longer there so he misses (even though on his screen it's a perfectly aimed, rested shot at what is basically a stationary target).

Eventually the low-ping player is going to get lucky. The high-ping player simply can't get lucky; his ping means that he's always shooting at a target that exists only for his client and not for the server.

In reality the outcome would be the opposite. The soldier firing from the rested, stable position is more accurate and his reaction time + bullet flight time is well within the time his opponent needs to lean out from cover and snap off a shot. The second time the guy leans out, when his opponent knows where to expect him, he's dead.

Due solely to ping, the opposite can easily occur in game. That's a pretty awful outcome for a realistic shooter.

 

[Serial Messiah]

It sounds to me like a lot of people are confusing client-side hit detection with client-side prediction and server-side lag compensation. Please keep your terms straight to avoid confusion.

Now for a good example of the latter, see any Valve game. I might be biased after playing CS and various Source mods for ages, but I very much prefer it over the kind of delays we are seeing with the current system in RO2.

I'm a little late to respond, but Source of all things. Source? Every Source game has somewhere between mediocre and awful hit detection. TF2's is mediocre, while CS:S and DoD:S are awful and L4D and to a slightly lesser extent L4D2 are horribly awful abysmal messes where you shoot some piece of **** four times in the head and nothing happens. You also have the times you get pegged from around corners. You know, usually between half a second to a full second after you reach cover. Source is a horrible example of hit detection or net code done right. BF3 is much the same, although the hit detection isn't quite as bad and the cover-negating corner-shooting lag bullet garbage is worse. Whereas every Unreal engine game I've played within memory has solid net code and certainly far better than any Source or Gold Source game or BF3.

I could understand some lag compensation in this, but it should be minimal and optional if it is implemented at all. Lag compensation is a bit of a dinosaur and a lot of games perpetuate it despite the obvious and frustrating flaws.

 

[Temporary]

excellent animated gif

I too wished Tripwire was actually working on fixing the game and listening to the players...

apparently they don't really care if Red Orchestra 2 is dying, they already have the money from preorder and the Christmas sales, why would they bother ?

 

[Comrade Kaizer]

excellent animated gif

I too wished Tripwire was actually working on fixing the game and listening to the players...

apparently they don't really care if Red Orchestra 2 is dying, they already have the money from preorder and the Christmas sales, why would they bother ?

As displeased as I was with RO2 post release, I wouldn't say that is necessarily the case here...

As some of you may have noticed in our steam OGG messages, we are playing around with the servers to test various things related to players reports on high player count/ping issues. This work will continue in the near future with more requests but we are putting pen to pad right now with what we've learned.

 

[Reise]

"Zero Ping" mutators did exist for UT.

They were mainly used on instagib/zoom instagib and the infamous sniper maps, where all players sat around with a sniper rifle and played point-and-click until their hands fell off.

Nobody in their right mind used ZP on normal matches. ZP instagib was also incredibly dodgy because you could not trust your own movements. Well, movement in general was pointless. You could bounce around for a while but someone would ALWAYS hit you. It was never as good as a nice standard low-ping server.

It was also heavily prone to aimbot abuse. Client-side detection and all that. Hell I even played with one a little bit for ****s and giggles. All hitscan guns worked with ZP aimbots.

Consider that for a bit.

 

[Mekhazzio]

ZP instagib was also incredibly dodgy because you could not trust your own movements. Well, movement in general was pointless. You could bounce around for a while but someone would ALWAYS hit you.

That sounds like it was working as intended, to me. A perfectly accurate hitscan weapon is supposed to be easy to hit with and impossible to dodge, that's the whole point of them. The networking in both Quake and Unreal gave you some wiggle room with the railguns with the usual 100ms online, but that's an error rather than an intended part of the design. On a LAN, those weapons devolved the game into point & click adventures. I've always felt it was poor game design to include them in action shooter games in the first place, with instagib just being the final evolution of a bad idea.

But yeah, I'm not sure how secure the mutator approach is. I was more of a Quake guy myself, I've never really dug into the nuts and bolts of UE before now. If mutator script can be gotten at without modifying the client process, that would be a Bad Thing, and all the more reason that Tripwire needs to do it on their side of things, but if it still takes process modification, then meh, that's no different than the bog standard aimbots.

 

[Reise]

Of course it was working as intended. It essentially band-aided the problem of having to compensate for latency.

The issue, which is at the root of the entire discussion, is that neither system is perfect. It is my opinion that the only way to solve it is with conventional approaches. Working out netcode issues, and getting to a point where you have server performance like those you see with the dedication in CS. Nice, local, sub-30ms servers, or lower.

With a compensation band-aid, you can definitely depend on your shots hitting where you point them. The trade-off is now you have to predict where the server is going to put you when the other guy pulls the trigger, instead of the guy pulling the trigger predicting where his shot is going to land. And with that system, improving netcode and server-client latency will only improve your predictions on where the server is going to decide where you die.

It's an argument of philosophy. Is the shot more important than tactical movements? Or vice versa? If you ask me, it is FAR easier to predict where your shot will land than to predict where the server has you pinned when the other guy pulls the trigger. THAT is why I will never support client-side hit registration or compensation. With shot leading, and accurate player positioning, you at least have the ability to accurately react in a firefight. When that foresight is removed via compensation, that ability to react is gone, and you are further removed from the game for it.

Your position ends up being at the mercy of the client, and there's no way to lead for that. Compensation has its place, but that place is not in any shooter that aims to be competitive.

 

[Mekhazzio]

Working out netcode issues, and getting to a point where you have server performance like those you see with the dedication in CS. Nice, local, sub-30ms servers, or lower.

I don't see this ever happening with RO, even if the server performance becomes completely transparent (and I've looked at the script in the SDK, that would take divine intervention. It's a mess in there)

We've got something like a few hundred players active in North America at any one time. That's a few powers of magnitude off from the quantity you need to get enough geographic saturation for just a majority to be able to play on local servers. "Oh, just play it on servers you get 30ms to" is a viable, if a bit elitist, philosophy for a hugely popular title, but it's a wishful fantasy for a niche game. RO will never be the sort of blockbuster that can get away with that.

you at least have the ability to accurately react in a firefight. When that foresight is removed via compensation, that ability to react is gone

Honest question here....what are you reacting to, and in what way? Given the fairly sedate movement speeds in RO2, the inertia on top of it, and the action time even on things like canceling a lean, it looks to my view like there's virtually nothing a person can do to escape their fate if they weren't already planning well ahead for it.

I mean, sure, we all know the games you can play with the networking, the zig-zags and the abrupt stops and the peeking too fast for anyone to hit you without psychic powers, but if you take all that away, say we're playing on a LAN connection with a monster server, what's left to react with in RO?

 

[Reise]

Honest question here....what are you reacting to, and in what way? Given the fairly sedate movement speeds in RO2, the inertia on top of it, and the action time even on things like canceling a lean, it looks to my view like there's virtually nothing a person can do to escape their fate if they weren't already planning well ahead for it.

I mean, sure, we all know the games you can play with the networking, the zig-zags and the abrupt stops and the peeking too fast for anyone to hit you without psychic powers, but if you take all that away, say we're playing on a LAN connection with a monster server, what's left to react with in RO?

I could ask the same thing about the way it is now. If RO2 was played on a LAN, you'd have very little to no leading to begin with. Especially with a dedicated server.

In RO I'd say those "reactions" are best applied to the smaller situations. Those times when you notice the enemy must be reloading, or when others may have run past and drawn their attention. You need to be able to have confidence that if you DO happen to make that step, that you can rely on it. Whole decisions on movement and progression rely on your ability to trust that where you'll be is the gold standard as far as the server is concerned. With the alternative, there is no secondary input (like leading) that you can use to correct it if your location happens to be off a bit.

In RO I fear compensation would cause what few players remain to lose that confidence and prefer the safety of cover. After being shot down callously by enemies whom the server pays the most attention to, players would end up sitting still and trying to make their own shots instead. It would lend itself to a situation where assaults just flat out don't happen anymore. With RO's slow pacing it would absolutely crush anyone's intentions of getting across that street or through those lines without getting utterly annihilated.

Would they have anyway? Nobody can honestly say. And that leads back to the philosophy of it.

 

[Rehmes]

This is a major deal. I'm slightly tired of feeling like im shooting fast moving tanks with a rocket when I'm really just shooting soldiers running. Shooting way ahead of them and boom headshot. It feels slightly off.

 

[Mekhazzio]

Those times when you notice the enemy must be reloading, or when others may have run past and drawn their attention. You need to be able to have confidence that if you DO happen to make that step, that you can rely on it. Whole decisions on movement and progression rely on your ability to trust that where you'll be is the gold standard as far as the server is concerned.

I'm just not seeing how even the worst-case scenario of a ~200ms lag time behind your client's position estimate is a make or break on any of that. Those aren't reactive decisions, they're simply choosing an execution time for a plan you've already developed, and they're based on fuzzy data to begin with. A situation where you can estimate timing to within a fifth of a second's precision for both your own movement and your window of opportunity is going to be rare, if for no other reason than the heavy fog of war RO has regarding the enemy.

My goal in bringing up the hypothetical LAN play comparison was to see if you have an edge case, a scenario that, from the target's persective, could play out differently online with a latency-compensated model than it would play out in a theoretically pristine environment. My "dying behind cover" video back there, for instance, would run exactly the same if I'd done that on a LAN. I would still be dead because I crossed in front of the window, I'd just get to hear about it sooner. While it's a flaw resulting from the system design, it's a flaw that doesn't have a gameplay impact.

In RO I fear compensation would cause what few players remain to lose that confidence and prefer the safety of cover.

The confidence that comes of knowing the artificial safety that movement confers? I'd happily be rid of that. Gameplay design points like this, the level of risk desired in movement, are things that should be controlled through gameplay solutions, not in a roundabout manner through technical oddities. The oddities can throw in shifting, arbitrary and often unknowable changes. If the goal is to make it, say, 80% safe to sprint across a hallway that a rifleman is covering, then the goal should be to make that consistently happen, and not fudge your way to it through a networking design that means you're 100% safe against most people, 0% safe against bots, and 40% safe against those two guys who live next door to the server.

 

[Gaizokubanou]

It's an argument of philosophy. Is the shot more important than tactical movements? Or vice versa? If you ask me, it is FAR easier to predict where your shot will land than to predict where the server has you pinned when the other guy pulls the trigger. THAT is why I will never support client-side hit registration or compensation. With shot leading, and accurate player positioning, you at least have the ability to accurately react in a firefight. When that foresight is removed via compensation, that ability to react is gone, and you are further removed from the game for it.

Tactical movement, no matter how good, will fail against an aimbot. Aimbot is the most extreme example of a good shooter, so good that it's basically not a fair play... but my contention is this; tactical movement in it's purest form is never meant as absolute defense against good shots. Both real life WWII combat and FPS gaming in general works around the principle that offense > defense. There is no absolute defense, because most small arms require line of sight (which means at the very least your head will be exposed during attacking). The best you can do is buy yourself enough time so that you can come up with your own offense to neutralize your enemy. You make yourself as small of a target as possible, so that it'll take too long for your enemy to take a good shot at you.

That's what good tactical movement does. It's not a guarantee that you will live, because that's really ultimately up to your enemy the moment you enter his/her vision. The moment the tiniest part of you is revealed, your life really hangs on your opponent's aim. But your opponents are human, so the idea is that with good enough tactical movement, you will make it hard for them to make good use of that slight vulnerability you have revealed in order for you to gain further advantage. In the end, you are making a gamble that your opponent is simply not good enough to take you out while you move to more favorable spot.

Aiming with a gun, on the other hand, is much more direct. While it's not absolute due to bullet drop, wind, and barrel fitting, the point is that outcome is much more obvious. You point it, and bullet travels at certain speed. The expectation that the shooter has of his/her own shooting has to be spot on because It's no one else but you. Unlike tactical movement where you have an uncontrollable variable of your opponent's aim, in shooting, unless your target is fast enough to actually actively dodge bullet, it all boils down to your own aim.

This is why I believe that sacrificing some precision in tactical movement is better choice than sacrificing aiming.

 

[RJ_MacReady]

I've been playing Grain Elevator last week and have been occupying one of the upper floors as a (lvl 50) MN rifleman. The Jerries attacked the main building and begun to climb the main stairwell. I've noticed two of them and shot one soldier dead from above. The second one began spraying the area in front of him not knowing where the shooter (me) was. I shot at him, aiming perfectly at his body volume... Missed. He crouch-strafed. I shot again (perfectly aligned sights!)... missed for the second time. Shot for the third time... missed. I emptied the whole MN magazine and all I managed to hit was a concrete floor by his feet.
Hookay... I reloaded. I shot another bullet towards the enemy. Missed. He strafed (Do I have to mention he wasn't aware of my position and my ping was ~90?). I just couldn't bloody shoot the guy! I felt like someone put me in a Matrix world.
Then suddenly he looked up and BAM BAM BAM! I died. How about that?