Leading for ping is ridiculous.
- Thread starter Synesthesia
- Start date
Well, there should be one thread for one problem. Not multiple threads.
Pretty much everything with a q3 engine gets it right, like cod4 of course the hitboxes where huge.
Dont confuse things, you even have to lead ahead with a ~50 ping which translates to an even better ping when you use the other game's method to calculate it.
This happens on excellent 32p servers that are a few 100km away and where nobody has a high ping.
Dont confuse things, you even have to lead ahead with a ~50 ping which translates to an even better ping when you use the other game's method to calculate it.
Last edited:
I've never been a Battlefield fan, so I've no idea what they're like these days, but let's see... the poster child is pretty much everything Source based, that covers from CSS to L4D to TF2 and even the glorified mods like Dino D-Day get to inherit Source's quality latency compensation.
You're not stuck with the engine default, however, because it's generally not a complicated feature to add. Both Monday Night Combat (Unreal) and Brink (idtech) strapped their own client-side hit detection onto a basic engine that didn't have it by default, and they both handle much more dynamic combat than RO2 will ever have. Serious Sam 3 does also, and of course, every single multiplayer flight sim ever made from IL-2 all the way back to Air Warrior. Even that new Aliens vs Predator game, for which multiplayer was a neglected afterthought, has client-side hit detection.
The trick really is finding a FPS that -doesn't- have latency compensation. You limit it to "get it right" but really, it's a simple concept that's fairly hard to screw up algorithmically. If BC2 is really that bad at it, they'd be the exception rather than the rule.
Last edited:
I don't think you understand what's being discussed in this thread. That L4D behavior is not at all relevant. The hitbox displacement is not related to the networking in any way and it's also not "iffy", it's intentional game design. The Source games forward displace the hit boxes on moving targets to approximate ballistics lead with hitscan weapons. Without the displacement, you would aim directly at the leaping hunters, which is unnatural enough to completely throw off anyone expecting a semblance of real-world weapon behavior. Think clay pigeons. It's fairly unique to Valve's games. Most shooters with hitscan weapons don't do it, and it's got nothing to do with how the networking negotiates hits.
You might be confused in that it involves adding lead, but it's critically different from the RO networking issue discussed here because it is not a factor of latency. In an online environment, L4D corrects for your latency so, as long as you were firing at the hitbox, the shot will hit, period. The hitbox displacement works the same for everyone, be it in single player, LAN or online, and it's always by the same amount. Once you know how to shoot a pouncing hunter, that's how you do it, consistently, all the time. It's completely up to the shooter.
In RO, by contrast, the networking-added lead is purely a factor of an ever-fluctuating network delay, and it's in addition to the actual weapon ballistics. To describe it visually, imagine that L4D Hunter was in RO2 and he didn't have the hitbox displacement. The hitboxes displayed in that video would be roughly where you would want to aim with a PPSh in order to hit him, to account for the bullet flight time. But that would only hit him if you were in single player. If you were online with RO2, to actually hit the Hunter, you have to aim even farther ahead by some amount based on your latency (a decent connection would be about 7 to 10 times farther ahead than the hitbox is) and that extra distance would fluctuate constantly based on internet connection vagaries and how your actions line up with server tickrate. Additionally, if the hunter were to change direction between when you fire and when your message gets to the server, then even if you were aiming in what was once the right place, your shot would still miss. Since you can't predict an aimpoint with any sort of reliability, there's now a large element of random chance introduced and it's largely out of the shooter's hands. A Hunter would be nearly impossible to hit in RO2's networking model.
Essentially, it's the difference between shooting bullets at a target and lobbing trigger-delayed paintballs at them.
You might be confused in that it involves adding lead, but it's critically different from the RO networking issue discussed here because it is not a factor of latency. In an online environment, L4D corrects for your latency so, as long as you were firing at the hitbox, the shot will hit, period. The hitbox displacement works the same for everyone, be it in single player, LAN or online, and it's always by the same amount. Once you know how to shoot a pouncing hunter, that's how you do it, consistently, all the time. It's completely up to the shooter.
In RO, by contrast, the networking-added lead is purely a factor of an ever-fluctuating network delay, and it's in addition to the actual weapon ballistics. To describe it visually, imagine that L4D Hunter was in RO2 and he didn't have the hitbox displacement. The hitboxes displayed in that video would be roughly where you would want to aim with a PPSh in order to hit him, to account for the bullet flight time. But that would only hit him if you were in single player. If you were online with RO2, to actually hit the Hunter, you have to aim even farther ahead by some amount based on your latency (a decent connection would be about 7 to 10 times farther ahead than the hitbox is) and that extra distance would fluctuate constantly based on internet connection vagaries and how your actions line up with server tickrate. Additionally, if the hunter were to change direction between when you fire and when your message gets to the server, then even if you were aiming in what was once the right place, your shot would still miss. Since you can't predict an aimpoint with any sort of reliability, there's now a large element of random chance introduced and it's largely out of the shooter's hands. A Hunter would be nearly impossible to hit in RO2's networking model.
Essentially, it's the difference between shooting bullets at a target and lobbing trigger-delayed paintballs at them.
Last edited:
Mehkazzio, thoughtful and well spoken posts. Thank you. I do think that some form of ping compensation or client side hit detection should be added, I had no real problem with BC2, apart from the occaisonal "Oh wow, I thought I made it into cover", compared to having to lead *every shot*. There is a reason big companies use client side/ping compensation, it is better.
Good post, Mekhazzio.
Question: is it possible to have such a client based system based on physics, or is everything then just glorified hit-scan? Do we sacrifice bullet drop and bullet travel time for the sake of such a system? I'd hate to think we have to give up proper ballistic behavior to correct lead-times.
As much as I hesitate to mention the A word as it might 'rouse the rabble, how does ArmA do it? I only ask because it's the closest comparable game I can think of that might use real projectile physics. Personally I don't want to give up drop and velocity just for the sake of high pings. It's one of the things that makes RO special. Even Rainbow Six and its many, many children use their expanding cone thing, which is just another twist on hitscan. Six feet or six hundred, the trajectory is the same (up to Raven Shield I should say, didn't play beyond that).
I only ask as a genuine query, not to be an arse.
Question: is it possible to have such a client based system based on physics, or is everything then just glorified hit-scan? Do we sacrifice bullet drop and bullet travel time for the sake of such a system? I'd hate to think we have to give up proper ballistic behavior to correct lead-times.
As much as I hesitate to mention the A word as it might 'rouse the rabble, how does ArmA do it? I only ask because it's the closest comparable game I can think of that might use real projectile physics. Personally I don't want to give up drop and velocity just for the sake of high pings. It's one of the things that makes RO special. Even Rainbow Six and its many, many children use their expanding cone thing, which is just another twist on hitscan. Six feet or six hundred, the trajectory is the same (up to Raven Shield I should say, didn't play beyond that).
I only ask as a genuine query, not to be an arse.
Ballistics don't have to be sacrificed. How the networking negotiates hits is independent of how the hits are computed. The algorithm outline goes:
- Client decides if it hits
- Client tells server that it got a hit
- (optional) Server decides if it wants to accept or reject that hit
- Server processes the outcome
This segues into another advantage of the method. Because the clients are calculating their own ballistics, the server no longer has to do it. That takes a significant workload off the server and distributes it across all the clients. Servers overloaded by a player count higher than they can handle are still a common problem, so that'd be of benefit on its own.
so i kind of skipped over a lot of this thread, but if you want to get used to the "aim ahead" mechanic, just play on dialup.
i play on dialup cause i'm forced to, and it took some getting used to knowing how far ahead to aim to compensate for 500-700 ping.
i play on dialup cause i'm forced to, and it took some getting used to knowing how far ahead to aim to compensate for 500-700 ping.
Definitely would prefer what folks are calling client-side detection. Having played lots of games that use it for thousands of hours, I find it much preferable to the insane bullet travel time in RO2. Only on rare occasions did I experience getting "killed behind cover," really the only infuriating aspect of it is when you shoot another player at the same exact moment he shoots you and one of you survives when it should've been a double KO. But even that situation is rare.
Now what we have now is a shooting system that is downright frustrating, in a game about shooting that is supposed to handle weapons realistically. Sure, you can get used to leading targets - assuming you know exactly what your ping is at all times in addition to the distance to your victim. Up close firefights just feel wonky unless you allow them to walk into your sights, a tactic I feel is necessary a lot of times in big games where the best ping you can hope for is 150ms. I don't even know why people bother hosting servers that accept more than 36, 40 players tops.
Don't get me wrong, RO2 is a great game and one of my favorites, but you have to acknowledge this is a problem.
Now what we have now is a shooting system that is downright frustrating, in a game about shooting that is supposed to handle weapons realistically. Sure, you can get used to leading targets - assuming you know exactly what your ping is at all times in addition to the distance to your victim. Up close firefights just feel wonky unless you allow them to walk into your sights, a tactic I feel is necessary a lot of times in big games where the best ping you can hope for is 150ms. I don't even know why people bother hosting servers that accept more than 36, 40 players tops.
Don't get me wrong, RO2 is a great game and one of my favorites, but you have to acknowledge this is a problem.
Did you really have to bring this one up again? 
Client-side hit detection is not the saviour of MP games, and this is quite evident when scanning the interwebs for discussion of this very topic. Each approach has pros and cons, and the client-side approaches prioritise either the shooter or the target - this is far from an ideal situation, even if it does improve "somewhat" on the server detection we have now. This improvement is entirely subjective, too, and biased depending on your particular opinion or annoyance
Bottom line is - TWI are unlikely to change the entire networking model, and any code that is directly linked to it. Yes, they might do it - but I'm betting on the fact that they don't. Client-side detection is relatively easy - changing a system that expects server-side detection to one that works with client-side detection, would likely be a fair bit more work. At the least, TWI would have to go through the whole of the networking code, to ensure that it all still works as expected - but I'd wager it goes deeper still. RO2 is transferring and synchronising more data between client and server than most FPS games even need to bother about - it isn't going to be a simple change by any stretch of the imagination.
If TWI made this a switchable, or maybe per-server option, I'd love to give it a try and directly compare the models within the same game. I see a lot of people complaining about client-side detection in other, bigger (and simpler) games - so this is probably a fine example of not being able to please everybody, and accepting that your own opinions are biased towards a given solution for a particular reason
Client-side hit detection is not the saviour of MP games, and this is quite evident when scanning the interwebs for discussion of this very topic. Each approach has pros and cons, and the client-side approaches prioritise either the shooter or the target - this is far from an ideal situation, even if it does improve "somewhat" on the server detection we have now. This improvement is entirely subjective, too, and biased depending on your particular opinion or annoyance Bottom line is - TWI are unlikely to change the entire networking model, and any code that is directly linked to it. Yes, they might do it - but I'm betting on the fact that they don't. Client-side detection is relatively easy - changing a system that expects server-side detection to one that works with client-side detection, would likely be a fair bit more work. At the least, TWI would have to go through the whole of the networking code, to ensure that it all still works as expected - but I'd wager it goes deeper still. RO2 is transferring and synchronising more data between client and server than most FPS games even need to bother about - it isn't going to be a simple change by any stretch of the imagination.
If TWI made this a switchable, or maybe per-server option, I'd love to give it a try and directly compare the models within the same game. I see a lot of people complaining about client-side detection in other, bigger (and simpler) games - so this is probably a fine example of not being able to please everybody, and accepting that your own opinions are biased towards a given solution for a particular reason
People complained constantly about this in ROOST and nothing was ever changed, I wouldn't hold your breath.
I know this has been done to death, I dislike the current model because:
Ping is not stable, this isn't a crack at the game/code/servers, pings fluctuate naturally AFAIK and due to this the amount of lead changes practically with every shot.
Also, say a newbie comes in, plays campaign to get an understanding of the game, then goes online, everything they just 'learned' about the gun play changes dramatically even on a server close to them with a 'good' ping.
I really hope something changes, I feel I need to check the scoreboard before I even LDS so I can get a feel for how far ahead I should be aiming, does wonders for immersion that....
Ping is not stable, this isn't a crack at the game/code/servers, pings fluctuate naturally AFAIK and due to this the amount of lead changes practically with every shot.
Also, say a newbie comes in, plays campaign to get an understanding of the game, then goes online, everything they just 'learned' about the gun play changes dramatically even on a server close to them with a 'good' ping.
I really hope something changes, I feel I need to check the scoreboard before I even LDS so I can get a feel for how far ahead I should be aiming, does wonders for immersion that....