Networking, Lag Compensation, and Hit Detection

[corpsecreate]

In the case of COD, I've never seen it screen tear so I guess they would be duplicates which would mean you get judder. I'd much prefer judder over screen tearing

[wjang]

Feb 5, 2013 11:37:32 GMT -5 timeitself said:

Any idea why the ammo counter isn't in sync with the firing animation? Can we assign any significance to how long it takes to update?

i have the same question

[Dumien]

This is hugely visible in the zombie galvaknuckles spawn abuse during instakill. The ammo counter can look like it is a clip behind the shots if you are using doubletap and a fast firing weapon.

[toysrme]

i never noticed it different, but if it had changed IMO it would be to combat hacking an unlimited ammo bug/hack on the client end they likely found.
tho IDK why waste the limited resources on something like that when it aint gunna do anything when they host is hacking.

[lackingdamage]

Enjoyed reading that
Great post!

[pachiderm]

Wow, I am glad I found this forum. This explains a lot. May you all be blessed with non-ugly children.

[the1jeffy]

OK. I've seen this info presented a few different ways. There's still no definitive answer on where to go next. This lag comp issue went away during BLOPS1 for me, because I got a better connection. However, BLOPS2 seems worse, or at least not predictable.

So what, as and end user, can I do to minimize the issue and what should we be stressing as universal actions to take that will improve matchmaking and home networking so as to again minimize this.

Nothing bothers me more than seeing an enemy displaying the sprint animation then dead. Then watching the match from his perspective later and seeing that he had time to sprint, ads, then zorro swipe me with auto fire. That means I am so far behind the host (or maybe I am the host?) that those matches/lobbies are unplayable.

I have port forwarding set up. I have 'full cone' (i think) set up on my settings. Short of removing the router from the equation (which I cannot always do as my wife needs the internet, too). Am I missing something/ doing it wrong?

[bucket415]

My connection on COD is the same whether its just me using the internet or my wife is on netflix & my daughter's on the PC and my son is also playing COD on his xbox. 30 down and 4 up speeds.

[jukebox20]

For the ones saying something is wrong with Black Ops 2, you are not alone. Alot more people are complaining abou the lag/lag compensation/hit detection/broken camera in this game than any other Call of Duty game.



YOU DO NOT HAVE TO REWIRE YOUR HOUSE, BUY NEW ROUTERS, PORT FORWARD, UPGRADE YOUR INTERNET CONNECTION, DMZ, OR ANYTHING LIKE THAT! DO NOT LISTEN TO MOUSEY!

IT IS THE GAME!

I SAY AGAIN: DO NOT SPEND MONEY TO BETTER YOUR CONNECTION IN THIS GAME. IT WILL NOT WORK.

IF YOUR CONNECTIONS WERE FINE BEFORE PLAYING BLACK OPS 2, THEN IT IS THE GAME THAT'S MESSED UP, NOT YOUR CONNECTION!



Something is very wrong with this game and an it pisses me off to no end! I went back to MW2 for a bit to see if the connections were really as good as people say. I can confirm that it is true--my bullets hit where they should go, there were no more mind bullets/ jedi foresight/ bulldoo-doo kills against me, no more popping out around corners and killing me in what seems to be one or two bullets.

I've never felt such a great connection like MW2 in a good while. When you play Black Ops 2 for a while then play MW2 again, it feels like a relief/burden was lifted off your shoulders because your bullets hit the person. Whole different world.

Like I said, something is wrong with Black Ops 2 and YOU DO NOT NEED TO DO MORE THAN WHAT YOU'VE ALREADY HAD BEFORE BO2!

[jukebox20]

I can totally feel it!

Anyway, I was using a silenced UMP and was halfway across the map. It's not that I was able to kill him fast, which I didn't, but it was that bullets just landed where I aimed/saw it.

[the1jeffy]

Feb 18, 2013 9:29:04 GMT -5 Mousey said:

Well if going wired isn't an option then there isn't much you can do short of rewiring the house.

I mean its good to keep in mind that when multiple people are on the same internet connection, they share bandwidth. This usually wont be a big deal since cod is fairly conservative to accommodate people still on dial up. But its good to avoid streaming video/downloading/torrenting/other bandwidth heavy stuff at the same time.

Of course that wont exactly make your internet faster but it will reduce packet loss which is still very important (especially on slow connections).

Thanks for the response. I am wired. It just that my wife's tablet, our office PC and my phone all hit the router, too. She doesn't stream video while I play, or torrent, etc. She does play fb games, or surf mamabargains, and the like.

I guess what I am aiming toward is: Well then what.

Should we recommend hard wired, port forwarding, or other QoS as priorities, in that order? What I would like is to educate as many people on basic latency reduction, which makes matchmaking easier and better.

However, I have been reading some interesting theories about latency being built into all routed home networking, so it might be a crap shoot either way.

[Slick]

Even though it's probably the game, and the fact you're dealing with latancy due to playing on the net, sometimes with people thousands of miles away from you, don't rule out local causes for "lag".

Go wired for starters. Hook up the xbox with an ethernet cable for a couple days, just for the sake of trying it. Wireless is extremely undependable, and if you're even remotely depending on the connection to be stable, get off the wireless.

If you have multiple people in the house, but you should honestly do it anyway because who knows what autonomous being starts eating up bandwidth , learn how to set up a Quality of Service setting on your modem/router. It regulates the upload bandwidth on the lan, this either sets up a priority for your packets from the xbox going out, and /or dedicates an certain amount of bandwidth for the xbox. Unfortunately, there's no easy step by step. Start with your modem manufacture and then your model number.

After that's done, there's not much else left, but you can save precious milliseconds by switching to a wired controller, a smaller TV, and stuff, but these changes are very minor regarding latancy compared to the usual suspects.

None of these changes matter if your latancy or jitter or packet loss is bad period though. That's when your ISP comes into play. They have ownership of the line all the up to the outside of your house. So run speed tests, ping tests, make a record of daily performance values, there's so many options to try, I won't bother typing them all, just try the first two.
If you're not getting low jitter low latancy to somebody, you're paying for bad internet. Run these tests to a server within a few hundred miles of your location. I pull 25 millisecond ping to chicago from Ohio, and that's high compared to fiber and cable setups. Unfortunately, I'm stuck on dsl.

One more thing though: Your inside wiring could be killing you too, it can induce packet loss, on dsl at least, not sure about cable since it runs on cox cable rather than copper, but if you have dsl, run a homerun setup for testing if able. I ended up doing that rather than rewiring. You are responsible for the wiring in your house, so if that's causing issues, your ISP isn't responsible for it.

[the1jeffy]

Feb 19, 2013 12:07:59 GMT -5 Slick said:

Even though it's probably the game, and the fact you're dealing with latancy due to playing on the net, sometimes with people thousands of miles away from you, don't rule out local causes for "lag".

Go wired for starters. Hook up the xbox with an ethernet cable for a couple days, just for the sake of trying it. Wireless is extremely undependable, and if you're even remotely depending on the connection to be stable, get off the wireless.

If you have multiple people in the house, but you should honestly do it anyway because who knows what autonomous being starts eating up bandwidth , learn how to set up a Quality of Service setting on your modem/router. It regulates the upload bandwidth on the lan, this either sets up a priority for your packets from the xbox going out, and /or dedicates an certain about of bandwidth for the xbox. Unfortunately, there's no easy step by step. Start with your modem manufacture and then your model number.

After that's done, there's not much else left, but you can save precious milliseconds by switching to a wired controller, a smaller TV, and stuff, but these changes are very minor regarding latancy compared to the usual suspects.

None of these changes matter if your latancy or jitter or packet loss is bad period though. That's when your ISP comes into play. They have ownership of the line all the up to the outside of your house. So run speed tests, ping tests, make a record of daily performance values, there's so many options to try, I won't bother typing them all, just try the first two.
If you're not getting low jitter low latancy to somebody, your paying for bad internet. Run these tests to a server within a few hundred miles of your location. I pull 25 millisecond ping to chicago, and that's high compared to fiber and cable setups. Unfortunately, I'm stuck on dsl.

One more thing though: Your inside wiring could be killing you too, it can induce packet loss, on dsl at least, not sure about cable since it runs on cox cable rather than copper, but if you have dsl, run a homerun setup for testing if able. I ended up doing that rather than rewiring. You are responsible for the wiring in your house, so if that's causing issues, your ISP isn't responsible for it.

My inside cable wiring is all brand new in Nov 2011 when I bought my house. I had the wires run from the street and had the tech signal check them. All green.

My modem is a www.amazon.com/gp/product/B001UI2FPE/ref=oh_details_o02_s00_i00?ie=UTF8&psc=1

Better than average.

My router is www.dlink.com/us/en/home-solutions/connect/routers/dir-655-wireless-n-gigabit-router

I have it static IP'ed with port forwarding to the Xbox's IP and have the QoS stuff set to the same.

I will run the speed/ping test when I can, however, running them from my PC hardly tells the whole picture. And I've ran them before, when I first got the new install last year. I don't remember specifics, but I recall the pings were normal and very little jitter.

I will try direct modem to Xbox one of these nights and see what shakes loose.

Thanks for your responses. I appreciate it.

[otisman666]

Here is a question that I have been thinking about. The # of packets sent/received issue was discussed. I asked the question (and nobody knew) about if there is a LIMIT to what the consoles put on the developers for these things.

Now let's steer our thoughts towards the new consoles. If this # of packets sent/rec thing is one of the problems with these COD games and regardless of the capabilities of the new consoles if the console makers still have the same limits, then nothing is going to get better. These are just thoughts.

Does anyone know if any other games exceed the packet sent/received amount that COD uses?

[otisman666]

Mar 18, 2013 15:07:00 GMT -5 Mousey said:

Since the game uses listen servers, it would not be constrained by any bandwidth limitations set by microsoft or sony. It's certainly set by the CoD devs. It's probably set so low because CoD's huge playerbase would likely include plenty of people who have mutliple systems on the same router. If it was higher than those players would experience heavy packet loss.

Ah, now that is something I did not realize / understand. Hmm, interesting.

[Slick]

I'm still amazed this awesome thread isn't up there with the other stickies. I forgot about this thread entirely.

Feb 19, 2013 14:27:17 GMT -5 the1jeffy said:

I will run the speed/ping test when I can, however, running them from my PC hardly tells the whole picture. And I've ran them before, when I first got the new install last year. I don't remember specifics, but I recall the pings were normal and very little jitter.

I will try direct modem to Xbox one of these nights and see what shakes loose.

Thanks for your responses. I appreciate it.

You're welcome.

[Deleted]

Feb 9, 2013 19:20:06 GMT -5 Mousey said:

there was a vid a while back showing how the ammo counter is delayed more under worse lag, so you can safely assume the host keeps track of it.

I think you're right. In Theater Mode the ammo counter is perfectly in sync

[bucket415]

Based on the shit connection that plaques this game, I just listed my copy for sale on E-bay. Had enough. Should have done it a long time ago. I'm renting MW4 first. Hopefully BF4 with 64 players is epic on the next gen consoles. KZ3 might be good on next gen.

Now I only wish I had a way to transfer my season pass to someone....

[Zero IX]

Mar 20, 2013 8:17:56 GMT -5 bucket415 said:

Based on the doo-doo connection that plaques this game, I just listed my copy for sale on E-bay. Had enough. Should have done it a long time ago. I'm renting MW4 first. Hopefully BF4 with 64 players is epic on the next gen consoles. KZ3 might be good on next gen.

Now I only wish I had a way to transfer my season pass to someone....

License transfer probs

[Pentaza]

I have a question about host. For some reason, when I pull host, something seems to go wrong with the hit detection.

Let me explain what I mean. Suppose I'm stood still and there's an enemy running across in front of me, maybe 10 metres away, and I'm shooting at him. The bullets just don't seem to register properly, in the way they usually do when I'm not host. I find I almost cannot shoot anyone who is moving across my field of view; moving towards me or away from me is OK.

I'm pretty sure it's not latency, which shouldn't matter for hit detection anyway. I've played across the Atlantic with bros from this forum and, while initially expecting the game to be unplayable, in practise it's fine. Not perfect, but not terrible, either. It highlights the fact that the game seems to have been made with bad connections, rather than modern broadband connections, in mind.

Yes, I get some deaths that I could do nothing about, such as when some guy pops out from round a corner and I seem to die almost instantly, but we all know how that works. I'm probably at a latency disadvantage in very close-quarters head-on gunfights, but most engagements - at least with my play-style - are not like that, so it doesn't matter to me. The only networking issue I get generally is when there's excessive jitter or packet loss and players start moving around in a jerky manner, which makes them hard to shoot.

But back to the host disadvantage phenomenon. Artificial latency cannot account for the fact that players suddenly become difficult to shoot accurately. It feels (unscientific, unfortunately) like the hit box and visuals aren't lined up on screen. I can only assume this is either some bug, or some result of the console (PS3) processor being overloaded by poorly-written code.

Any thoughts?

[bucket415]

Host disadvantage exists.

[Pentaza]

Mar 30, 2013 12:52:44 GMT -5 Mousey said:

Mar 30, 2013 12:33:38 GMT -5 bucket415 said:

Host disadvantage exists.

Like this! People will just throw out a reason with no explanation whatsoever because heaven forbid the game doesn't perform exactly the same as an older one, people just come up with whatever explanation could even remotely fit the problem to help them sleep at night. And there's nothing wrong with that, but until someone goes so far as to even try to show a half decent test on the subject I'm just going to assume that whoever says the stuff hasn't either.

Unfortunately I don't have a capture card, otherwise I'd have a go at some proper testing, or at least capture some in-game footage so that I can play back in slow motion and see if I was shooting on target or not.

What I experience when host is definitely different to when not host and does appear to relate to hitbox issues, rather than network problems. And as I mentioned previously, I've played across the Atlantic on a 2/3-bar connection quite a few times, and not had the same issue as when I pull host against players in the same region as I am. I can tell the difference on a 2/3-bar because the netcode "features" (like getting shot round corners, etc.) happens slightly more often, but, irrelevant of the latency, "shoot where they are on your screen" still works fine.

And, to be clear about what I'm describing with host problems, it's not players moving jerkily across the screen, etc. Perfectly smooth, all looks fine. And it's not the shot round corner stuff, either. Just, very specifically, that bullets don't seem to land properly when I appear to be aiming on target in the usual way.

It may vary by platform, depending on the root cause (assuming the effect is real). For what it's worth, I have also noticed frame rate issues when host once or twice (i.e. hardly ever) and, specifically, on Turbine. So it could be something to do with the PS3 being overloaded when host.

If the effect can be proven, and is the same on both console platforms (may not happen on PC with the dedicated servers), then it's probably some bug.

Actually, this does raise a question about hit markers - I'm of the understanding that these are generated on the client, i.e. you see/hear them when the client decides that you've hit the target, and not having to wait for the server to acknowledge, first. Is this correct?

Another explanation for the host effect that I feel is that the game runs fine, correct hit detection, but a bug with the audio/visual hit marker feedback. I know I'd be particularly sensitive to this, because I really feel and rely on the hitmarker sound. I initially had trouble adjusting to BOII because of the soft and squishy hitmarker sound compared to MW3, which was a bit more sharp and meaty. And I use silenced weapons almost exclusively because I find the hit markers hard to feel (yes, I know, it's a sound!) when using unsilenced weapons.

[Pentaza]

Mar 30, 2013 14:37:05 GMT -5 Mousey said:

Well to be fair to test any artificial delays, you'd need to be on PC to see the ping value. Private matches use listen servers so it would be consistent. You'd need at least two players of course. You'd also have to compare it to compare it to an older one (CoD4 or W@W probably since they also show ping as a number).

Lastly, you'd have to find a way to synchronize the times of clients who are not on the same network, which is possible using the game's time limit. You'd only be able to see differences in it though, so you still wouldnt be able to draw a complete conclusion from it.

BUT ANYWAY i talked with marvel like two minutes ago and he's up for screwing around in this with me so sooner or later we may have something.

I think a good test for host hitbox issues would be to have one person (the host) ADS and firing a constant stream of bullets in a single direction, ideally with a fast-shooting weapon with large clip (one of the LMGs, perhaps) with a second player running through the stream of bullets at a 90 degree angle. Then play back in slow motion, work out which bullet(s) were credited with hitmarkers, and determine if those bullets were as the character was directly in the line of fire, or slightly before or slightly after.

For comparison purposes, the test would probably need to be reversed, with the non-host player firing the stream of bullets and the host running through, to see not only if the hitbox is accurate in both cases, but also to see if there is any difference.

If we break the result into three possible hypotheses...
1. There is no host hitbox problem, in the sense that the results are the same for the non-host.
2. There is a host hitbox problem, and it is out by a constant value, irrespective of other factors.
3. There is a host hitbox problem, and it is out based on some measurable factor, such as latency.
... Then it would probably be a good thing if there was a noticeable amount of latency between the host and other player, because if the third hypothesis was correct, it would make it easier to detect.

Just one other thing to throw into the mix: due to the small number of games which I ever find myself in a single lobby before moving to another, most of the time when I pull host, it's due to a mid-game migration. So if there is a host disadvantage, which I believe I've experienced, then it could be some bug related to host migration. This would be difficult to test.

[pachiderm]

Mar 30, 2013 19:08:46 GMT -5 Pentaza said:

Mar 30, 2013 14:37:05 GMT -5 Mousey said:

Well to be fair to test any artificial delays, you'd need to be on PC to see the ping value. Private matches use listen servers so it would be consistent. You'd need at least two players of course. You'd also have to compare it to compare it to an older one (CoD4 or W@W probably since they also show ping as a number).

Lastly, you'd have to find a way to synchronize the times of clients who are not on the same network, which is possible using the game's time limit. You'd only be able to see differences in it though, so you still wouldnt be able to draw a complete conclusion from it.

BUT ANYWAY i talked with marvel like two minutes ago and he's up for screwing around in this with me so sooner or later we may have something.

I think a good test for host hitbox issues would be to have one person (the host) ADS and firing a constant stream of bullets in a single direction, ideally with a fast-shooting weapon with large clip (one of the LMGs, perhaps) with a second player running through the stream of bullets at a 90 degree angle. Then play back in slow motion, work out which bullet(s) were credited with hitmarkers, and determine if those bullets were as the character was directly in the line of fire, or slightly before or slightly after.

For comparison purposes, the test would probably need to be reversed, with the non-host player firing the stream of bullets and the host running through, to see not only if the hitbox is accurate in both cases, but also to see if there is any difference.

If we break the result into three possible hypotheses...
1. There is no host hitbox problem, in the sense that the results are the same for the non-host.
2. There is a host hitbox problem, and it is out by a constant value, irrespective of other factors.
3. There is a host hitbox problem, and it is out based on some measurable factor, such as latency.
... Then it would probably be a good thing if there was a noticeable amount of latency between the host and other player, because if the third hypothesis was correct, it would make it easier to detect.

Just one other thing to throw into the mix: due to the small number of games which I ever find myself in a single lobby before moving to another, most of the time when I pull host, it's due to a mid-game migration. So if there is a host disadvantage, which I believe I've experienced, then it could be some bug related to host migration. This would be difficult to test.

I don't think the hitbox is the main problem with the game. As far as hit detection is concerned, Blops 2 is pretty good. The problem for me is player models not being accurately represented. I can't tell you how many times I've been shot by players that appear to be sprinting, or been killed by players who move impossibly fast through a door or around a corner and are somehow already shooting at me before I can even react.

[pachiderm]

Mar 30, 2013 14:19:17 GMT -5 Pentaza said:

Unfortunately I don't have a capture card, otherwise I'd have a go at some proper testing, or at least capture some in-game footage so that I can play back in slow motion and see if I was shooting on target or not.

Oven Baked Muffin did a video like that. He got a very interesting 160ms ping on LAN which doesn't make any sense at all. He also had some interesting thoughts about gameplay imbalances. Ignore the rest of the video though.

[Pentaza]

Mar 30, 2013 21:29:54 GMT -5 pachiderm said:

I don't think the hitbox is the main problem with the game. As far as hit detection is concerned, Blops 2 is pretty good. The problem for me is player models not being accurately represented. I can't tell you how many times I've been shot by players that appear to be sprinting, or been killed by players who move impossibly fast through a door or around a corner and are somehow already shooting at me before I can even react.

Note that I'm not claiming a general hitbox issue, just that something seems off when you pull host (and perhaps some other factors when host too, such as after a mid-game migration).

The other effects you mention sound normal to me and are explained by Mousey's write-up. They're "features" of the way the game runs. The best thing you can do is to adapt your play style to take advantage.

[Pentaza]

So I pulled host a couple of times recently, both as part of a migration (the easiest time to tell). I paid extra careful attention.

The first time, the hit detection did indeed seem off - like I was firing rubber bullets. After a couple of minutes I gave up and quit because I was struggling to hit anything at medium distance (most of my engagements).

The second time, the hit detection seemed totally spot on.

Unfortunately, even my subjective opinion is not observing consistent results.

[pachiderm]

Apr 2, 2013 17:14:34 GMT -5 Pentaza said:

So I pulled host a couple of times recently, both as part of a migration (the easiest time to tell). I paid extra careful attention.

The first time, the hit detection did indeed seem off - like I was firing rubber bullets. After a couple of minutes I gave up and quit because I was struggling to hit anything at medium distance (most of my engagements).

The second time, the hit detection seemed totally spot on.

Unfortunately, even my subjective opinion is not observing consistent results.

You aren't going to get consistent results. Lag works both ways. If you're host and everyone has a terrible connection to you, you'll have a harder time hitting people, just as they'll have a harder time hitting you. Likewise, If you're giving everyone a great connection you probably won't notice a problem with hit detection.

Most of the lag in CoD is the result of player's connections to each other, not some sort of weird aberration in the game's net code. I cringe whenever I hear youtubers yell at Treyarch and tell them to "fix the lag." Treyarch can't change everyone's connection, all they can do is improve the host selection, or adjust the lag compensation. I would, however, like an explanation on the 100+ ms delay on LAN though, or why they seem to have felt the need to add significant delay to everyone's connection in an attempt to make it more fair for players who are lagging their nuts off.

[bucket415]

When I am host I notice that I am .5 seconds behind everyone in the room. It takes 3x the bullets to kill one person and I die to anyone facing me instantly.

[asasa]

Apr 2, 2013 18:24:45 GMT -5 pachiderm said:

Apr 2, 2013 17:14:34 GMT -5 Pentaza said:

So I pulled host a couple of times recently, both as part of a migration (the easiest time to tell). I paid extra careful attention.

The first time, the hit detection did indeed seem off - like I was firing rubber bullets. After a couple of minutes I gave up and quit because I was struggling to hit anything at medium distance (most of my engagements).

The second time, the hit detection seemed totally spot on.

Unfortunately, even my subjective opinion is not observing consistent results.

You aren't going to get consistent results. Lag works both ways. If you're host and everyone has a terrible connection to you, you'll have a harder time hitting people, just as they'll have a harder time hitting you. Likewise, If you're giving everyone a great connection you probably won't notice a problem with hit detection.

Most of the lag in CoD is the result of player's connections to each other, not some sort of weird aberration in the game's net code. I cringe whenever I hear youtubers yell at Treyarch and tell them to "fix the lag." Treyarch can't change everyone's connection, all they can do is improve the host selection, or adjust the lag compensation. I would, however, like an explanation on the 100+ ms delay on LAN though, or why they seem to have felt the need to add significant delay to everyone's connection in an attempt to make it more fair for players who are lagging their nuts off.

Thats just not true. Hosting is almost always a death sentence, no matter what the ping of the other players in the lobby is.

If only everyone who hosted dashed every game. Maybe that would wake 3arc up and have them fix this shit. As for now, I'll continue that battle mostly alone.