Leading for ping is ridiculous.
- Thread starter Synesthesia
- Start date
honk if you want to beta test some client side hit detection!
It cant be that much of a code write up can it? (Honest question, i really dont know what kind of work does that take)
http://www.youtube.com/watch?feature=player_detailpage&v=DMynC4twXV8#t=19s
It cant be that much of a code write up can it? (Honest question, i really dont know what kind of work does that take)
http://www.youtube.com/watch?feature=player_detailpage&v=DMynC4twXV8#t=19s
i would volunteer! i'm a sucker for beta testing! makes me feel important for a few minutes! LOL!
In terms of game movement 200 to 300 milliseconds is quite a large distance traveled in some situations. Knowing where I die is often a huge clue to the situational awareness. If you get shot somewhere, you can think back and compare the angles that you might have been shot from.
So it might not be completely randomly, but the point is that it sure does feel that way.
I don't think I have to explain myself when I say that Red Orchestra 2 doesn't even come close to the noticability of such things compared to client side prediction games.
So you basically say that server side prediction with 150 ping still doesn't make for magical teleporting? The side effects of lag aren't just found in the delay from your end to the server. A lot of times, a higher ping also creates higher packet loss. These days developers code a lot of measures to prevent it from being too noticable. But all the side effects as well as the delay can make 150 ping very unplayable.
I think you really overexaggerate the effects of leading on low-ping gameplay. A body ingame does not have a straw-thin hitbox. With low ping you will hit regardless of whether you lead or not.
If the effects of lag bother you this much, then why would you want client side prediction in the first place. It is far more noticable than the lag found at 60 ping. Which is triple of what you find tolerable.
The tickrate is indeed quite low, but increasing it means that the server hardware will need to get better as well. This is just something that requires a few years to pass, and faster hardware will get cheaper. So in a couple of months or years, we might see an increase in double tickrate servers.
When I am behind a cover I expect to be safe from that direction. When it gets to me that I am supposed to be dead I don't know where the shoot came from and have no idea which covers are still safe and where the enemy might be. While it can be good for the shooter it is not for the shootee. Right now disadvantage is on the side of the shooter and making it client-side will only push it towards the shootee which can end in people being even less willing to advance or even move altogether. It's not a solution. It's more like throwing a rock from one garden into another.Synesthesia said:
The game's sprint speed is 5 meters per second. 300 milliseconds of that is 1.5 meters. It's about the length you take up when prone.
You think the current system works fine at a 60ms delay, which I disagree with, but we'll go with that for the sake of an example. I've just connected to the four most populated servers at the time I type this. The lowest ping any player had across all four servers was 74ms. Roughly a dozen players were under 100ms, the large majority were in the 120-180 range, and about a dozen were over 250. By your own metric, the system does not work for any of the players in the game right now, and most of them are beyond twice the threshold.
Sure, the game works passably well on an excellent connection, and flawlessly on a LAN, but that isn't relevant to how the game is being played. We can pretend that everyone in the world has a great ISP and a local server to play on, but the reality is completely different from that. Even if you're living in an area of better than average net infrastructure, RO is a niche game, and so the odds of having a populated server in close network proximity is, through simple geographical distribution, quite unlikely.
If a program cannot function correctly in its average test case, then its design is innately flawed. A calculator app that returns 1 + 1 = 4 should be considered a broken piece of software. What we have in RO2 is an online multiplayer shooter game that cannot deliver accurate shooting in an online multiplayer environment. That's something of a weakness.
just for the sake of illustration:
yeah, server side would do no harm to this median latency. *sigh*
why on earth would we revere this game's gunplay so much if we have to lead for our own latency? How can this be ever defended? Do we need videos of leading for cqc? I honestly dont get how this is not an issue for some people.
yeah, server side would do no harm to this median latency. *sigh*
why on earth would we revere this game's gunplay so much if we have to lead for our own latency? How can this be ever defended? Do we need videos of leading for cqc? I honestly dont get how this is not an issue for some people.
I don't think you really comprehend what I'm trying to say here do you. You take Red Orchestra 2 as an example, though what you're forgetting is that I never said that I found Red Orchestra 2 in a good state when it comes to ping.
If you've read my earlier posts, I clearly stated I found the ping people get in Valve games much more acceptable. And you really don't need an excellent connection or ISP to enjoy low pings in that game.
The remaining people that still have a high ping are those that come from another continent. There will always be players that play on servers that are not close to them, and they will still lag no matter what the method of lag compensation is.
Now I do know that in the current situation there are people that simply face a lack of servers close to them in some parts of the world. But that's simply an issue that's present with a smaller playerbase.
Once all the issues are sorted out, I am sure there will be hundreds and eventually a few thousand players that will come back to this game. I am pretty certain that you will see a reduction in high ping players on your local server.
The developers themselves admitted that full servers of let's say >50 people lagged more than they should. If you've played the beta, you may have seen the pre-punkbuster ping many had. Those were significantly lower than what we see to this day. They are looking into fixing those issues, so something like client side prediction is still entirely unneeded.
If you've read my earlier posts, I clearly stated I found the ping people get in Valve games much more acceptable. And you really don't need an excellent connection or ISP to enjoy low pings in that game.
This is even more true when considering client side prediction.
The remaining people that still have a high ping are those that come from another continent. There will always be players that play on servers that are not close to them, and they will still lag no matter what the method of lag compensation is.
Now I do know that in the current situation there are people that simply face a lack of servers close to them in some parts of the world. But that's simply an issue that's present with a smaller playerbase.
Once all the issues are sorted out, I am sure there will be hundreds and eventually a few thousand players that will come back to this game. I am pretty certain that you will see a reduction in high ping players on your local server.
The developers themselves admitted that full servers of let's say >50 people lagged more than they should. If you've played the beta, you may have seen the pre-punkbuster ping many had. Those were significantly lower than what we see to this day. They are looking into fixing those issues, so something like client side prediction is still entirely unneeded.
Last edited:
For the most part for me, not an issue. I am not saying others don't have a problem.
Also from that illustration, half the people on the server should be playing on other servers closer to where they live. If you choose to play on a server that gives you a higher ping, that is not the game's fault.
Also from that illustration, half the people on the server should be playing on other servers closer to where they live. If you choose to play on a server that gives you a higher ping, that is not the game's fault.
As I see it;
1) Client-side:
Pros:
+ Maintains integrity of the shooting experience at any (most?) latencies, high or low
+ Allows customers from countries lacking servers (very common, since RO2 is a very niche game) to play comfortably in international ones
Cons:
- Can distort the experience of being shot at ("shot behind cover")
- Easier to cheat I guess?
2) Server-side netcode ("UT Legacy netcode"):
Pros:
+ Maintains integrity of movement/map placement at any latencies
+ "Hit effects", being server side, are never "false positives"
Cons:
- Requires extremely low pings to get "correct" weapon behavior going (Even with 60ms latency you're getting an unrealistic 60ms VISUAL delay to your firing, and a 120ms VISUAL delay on seeing your shot hit the wall)
- Impossible (or nearly so) for clients without local servers to have much of a fighting chance in international servers, or much of any kind of enjoyable play
Given the choice, I'd go with client-side in a heartbeat. Keeping the guns and the shooting working fine for all players at the dubious expense of "getting killed behind walls" is a no-brainer for a first person shooter. With client-side hit detection, lower pings still have an undeniable advantage, but the differences are not the gaping chasm of the UT Legacy netcode.
Really, folks, why are we even using 90s netcode in 2011 still? Valve, IW, DICE and even id (who pioneered this network model) have already dropped it. It's not a console thing either - I got Serious Sam 3 on Steam on launch. It is PC-only and its netcode is client-side. I play on international servers with great ease, and this ease makes me want to keep playing.
---
1) Client-side:
Pros:
+ Maintains integrity of the shooting experience at any (most?) latencies, high or low
+ Allows customers from countries lacking servers (very common, since RO2 is a very niche game) to play comfortably in international ones
Cons:
- Can distort the experience of being shot at ("shot behind cover")
- Easier to cheat I guess?
2) Server-side netcode ("UT Legacy netcode"):
Pros:
+ Maintains integrity of movement/map placement at any latencies
+ "Hit effects", being server side, are never "false positives"
Cons:
- Requires extremely low pings to get "correct" weapon behavior going (Even with 60ms latency you're getting an unrealistic 60ms VISUAL delay to your firing, and a 120ms VISUAL delay on seeing your shot hit the wall)
- Impossible (or nearly so) for clients without local servers to have much of a fighting chance in international servers, or much of any kind of enjoyable play
Given the choice, I'd go with client-side in a heartbeat. Keeping the guns and the shooting working fine for all players at the dubious expense of "getting killed behind walls" is a no-brainer for a first person shooter. With client-side hit detection, lower pings still have an undeniable advantage, but the differences are not the gaping chasm of the UT Legacy netcode.
Really, folks, why are we even using 90s netcode in 2011 still? Valve, IW, DICE and even id (who pioneered this network model) have already dropped it. It's not a console thing either - I got Serious Sam 3 on Steam on launch. It is PC-only and its netcode is client-side. I play on international servers with great ease, and this ease makes me want to keep playing.
---
Kinda off topic, but the last time I played on the 7thCav server (which seems to have dissapered from my server list for some reason) an entire round of Pavlov's House had the entire server spike up to 400-900 ping, lowest in the server was around 350. Less serious spikes still frequently occur on any server with 40+ players.
The horrible bullet lag is the one thing stopping my friends and me from playing Red Orchestral 2.
It's just not fun playing.
It's just not fun playing.
I never said it is not an issue. I said it is lesser of two evils. Client-server hit detection is not the solution. It might help you and the people who don't like to lead targets, sure, but it will backlash at someone else. Like the people who don't like to die while being in cover. It will still be there, on someone's side and should be fixed entirely. Fixed. In situation where it can't be fixed the lesser evil should be enacted. Now we have to somehow define which is lesser and why... Both sides have their valid points.Synesthesia said:
A lot of people sure don't seem to know what client hit detection is. What many are talking about here is simply a different form of server hit detection. The client will never do any hit detection. EVER. For very obvious reasons.
This matter is further confused by the fact a depressing number of servers are hilariously badly configured. I'd say a great deal of servers are running more slots than they can handle. The moment the player number in such servers gets close to the max, everyone's ping goes up and up and up... until you get that classic situation of everyone showing 150-200 pings in F1.
This matter is further confused by the fact a depressing number of servers are hilariously badly configured. I'd say a great deal of servers are running more slots than they can handle. The moment the player number in such servers gets close to the max, everyone's ping goes up and up and up... until you get that classic situation of everyone showing 150-200 pings in F1.
qft.
So far, gunplay is just not fun. Holy.Death, i LOVE leading targets, the gunplay is what brought me to this game in the first place.
It is just not fun to play this as it is. Most people playing have relatively high pings, because red orchestra is a niche game, as was said, and this will stay this way. No steam sale will give us servers on southamerica, australia, or wherever. Believe me, we tried.
The gunplay is much more important than the cover thing, i know we can all get a bit frustrated when that happens, but ultimately, most of our deaths will be in the open, after our enemy has lined us up. If we can't shoot properly in a god damn tactical shooter, what is the point?
This is the only game of this generation i've seen with this sort of netcode. The pros and cons are very well put by mr Echo black up there. Let's fix this, please. A dev post would be nice.
Last edited:
You need a good ISP, and you need a player base large enough to support hundreds or thousands of active servers so that you have good odds of finding one in close network proximity. Also, you seem to not be aware that Valve has had latency compensation in their games for a decade. They spend three paragraphs explaining why they think the "died around a corner" phenomenon is unimportant, including this gem:
Of course their games play better online, that's because their software is designed better. That's the point we've been trying to make in these threads.Valve Software said:
The current networking model is one of those issues that is keeping those people away. Look at player reviews, professional reviews, in-game youtube videos, etc. There are so many complaints about how shooting is simply not working as it should everywhere else. These forums are the only place I've seen people cheering the server-side, non-compensated hit detection as a good thing -- side-by-side with threads complaining that hit-registration is bugged, which can be neatly tracked down to... The server-side non-compensated hit detection.
You forgot:
- Can distort the experience of being shot at ("shot behind cover")
The people advocating the server-side non-compensated model like to overlook the fact that it, too, results in you dying when you appear behind cover on your screen. When you die on the server, you still don't have the message on your end. When you finally get it, you drop dead, even if you've already moved behind cover on your computer. It's a "problem" (And an illusionary one at that) that exists in either model.
As for my own summary:
You can make the networking work for only one of three perspectives; the shooter, the server, or the target. Only one of these has you shoot where you actually should, same as if you were playing a single-player or LAN game. Only one of these has no "dying behind cover" (And it's a horrible model, more on that in a sec). And one of these, the server, is unseen by any of the players.
To get rid of the illusion of dying behind cover, you would need target-side hit detection. Now you not only have to lead by your own latency, you have to lead by the target's latency. Quick, is that sprinting guy you're shooting at pushing 100ms or 200ms latency? That's a half-meter difference in where you have to aim, in addition to whatever yours is, plus whatever lead you actually need for his movement. Good luck hitting anything moving with a bolt-action. Especially if he changes direction. There's a reason this model was dropped, and dropped hard.
So that leaves us with modeling either for the server's perspective, or the shooter's perspective. And here I think people are very much confusing what is actually happening. See, when people complain about "dying behind cover," the fact is, they're not. It looks like it, but that's an illusion created by the simple fact that when you play a game over the internet, there is going to be a delay between all the players and the server. In fact, you did not die behind cover. That open area you just moved through a fraction of a second earlier? That's where you died. This happens in client-side, server-side, and server-side lag-compensated models. Yet despite how horrible this dying-behind-cover deal is, there's a notable silence about the event at the moment, despite it happening.
But "dying behind cover" is nothing but an illusion. You are not any more vulnerable to an accurate shot in either model, and you are not exposed to fire any longer. People seem to treat the client-side model as skewing this. If we're dealing with, say, 200ms latency, a client-side model does not mean that you're exposed for 200ms longer, rather you're exposed 200ms later. If you sprint across a 10-meter gap, you're exposed for the same 2 seconds either way, and if someone shoots in the right place, you die in the same spot, either way.
For a more thorough example: Say you're sprinting along when you enter an area covered by someone with absurdly high latency (1,000ms) just 20 meters away. With a server-side model, you enter his sight. Because the server predicts movement he sees where the server says you are. Let's say he's on the ball, and in half a second (Distance traveled 2.5m) not only can estimate exactly how far he has to lead you (About 5 meters, or nearly 3 full body lengths) but can sight in and shoot. He does, and one second later (Total distance traveled, 7.5m), that information hits the server. Server compares data, says, "yup, he's dead," and pushes that data to you. You drop dead 7.5 meters out.
Now for the client-side model. You enter his sight. He again takes a half-second to get a bead on you and fire (2.5m). His client says "It's a hit!" and sends that to the server. It arrives a second later (7.5m). Server looks it over, agrees that the shot could have been valid, declares "yup, he's dead," and pushes that data to you. You drop dead 7.5 meters out.
Not a very big difference there, is it?
Oh, but you can find places where it does make a difference, and let's just say those are not so good. People complain constantly of the run-and-gun tactics, yet we're ignoring that the server-side hit-detection is possibly the largest factor contributing to that. Hitting a running target is far harder than it should be, and hitting one that is moving evasively is even more so, especially for bolt-action rifles. At 20 meters, a target's movement should not matter much (10-20cm, perhaps), but in the current model, it can mean over a meter of lead needed, and if they're maneuvering, you have to guess what way they're going. Ever had someone hip-firing a SMG or MG while running zig-zag at you, while you're prone and trying to hit them with your bolt-action rifle aimed down the hallway they're charging up? Good money on which way that one's going to end.
With the current model, you are ridiculously easier to hit while stationary than when you're moving. Poke your head over a fence at 150 meters in Spartanovka, you've probably going to lose it in short order. Run across the huge field to the gullies at 150 meters, and you'll probably be just fine.
A networking model that ruins the firearms handling for anyone without a very low ping, in a niche game that is unlikely to have servers nearby for everyone, is a serious flaw. There is a very good reason that this model has been dropped for practically every FPS released in the past decade.
That's your own interpretation. From what I can read, they have opted for a hybrid model that still does server checks in order to prevent cheaters.
That does not magically mean that they find the "died around a corner" phenomenon to be unimportant. But it just means that the ping most people have on Valve servers is low enough to make a hybrid model possible without experiencing many of the issues it has.
If Valve games would have a similar ping to Red Orchestra 2, then I can guarantee you that even their system can't keep up.
It's a somewhat technical document, so I suppose some misunderstanding of it is to be expected. The Valve system is full-on latency compensation, not some sort of compromise. The server keeps a short-term record of history, and when someone with a 400ms delay takes a shot, the server looks through its history for a game state 400ms in the past and checks the shot on that game state, not the current one. Checking shot validation against past events is exactly the same thing that client-side hit detection does, Valve's method simply has it being performed on the server instead of a client to attempt to bypass any issues with untrustable clients.
What are you offering for this guarantee? I'll take you up on it, because it'll take only about 20 minutes to record good game footage to post on youtube to demonstrate how incorrect that assumption is.
Over the last two or three years, I've been playing TF2 on servers that I get a 150-200ms ping to, because I prefer the community there over picking just whatever random public server I get a great connection to. I've got a couple hundred hours on the sniper class in that game, so I've spent a fair bit of time making, and consistently landing, precision shots against targets that are moving rapidly and erratically. Ping time has basically no effect on the difficulty of those shots, thanks to Valve's latency compensation. You still aim directly at the enemy to hit them, no matter what your ping is.
Contrast this with Red Orchestra, where you have to lead every target all the time. Without latency compensation, even the slow ironsighted sidestep is enough motion to make directly aimed shots miss completely.
The difference between the two games is marked, and it has nothing to do with a ping difference. I actually get a better average ping to the 2.FJg server I play RO2 on than I get to the TF2 community server.