Weird isn't pleased.

I pretty much stopped playing because I couldn't stand the way the gamestyles were developing as a result of the netcode....

So ye you could say that I'm not pleased... but I warned you guys about this years ago. And I warned ID about it about a year before I posted anything publical.

is "publical" correct english?

I somewhat doubt it :D
I guess it should be publically or public.

But my view on language is that: If someone can tell me how it should've been written. The message came across just fine.

i almost agreed with u till u somehow lost the second part of the 'if' sentence. if A, then B. A, hence B
if (A)someone knows what the correct language is, then (B) it's ok even if he makes mistakes (due to laziness or whatever)

First of all it's dangerous to even think about agreeing with me....

Secondly I didn't lose any part of the if.
You're mis-reading my statement.

For my A to hold it doesn't matter if the writer knows the correct spelling. It only matters if the reader understands the message.

If the reader, who knows how to write, understands the message clear enough such that he can accurately tell the writer how it should've been written. ==> It's no problem at all as the purpose of language (to convey a message) is perserved.

yeah but what u just described is peasant shit, dumbass cracker shit. an ignorant hick has some expressiveness or gets lucky and the arguably educated reader understands what he meant. that aint kosher in my book. it's like getting lucky. or even worse, it's like u owe the existence of the discussion to your educated partner (cuz if 2 dumbasses would only end up killin each other before they'd come close to communicating and getting the message across). and being in any sort of debt is something i personally avoid at all costs. at least if u have any personal standards or a grain of respect for language and elementary general culture
or... kurwa whatever ur an engineer or programmer or some shit like that yall are. it's like God talkin to insects lal why do i bother. well, i dont actually im stalling getting back to studying the shit for the exam tomorrow.. today.. 3 hours in fact holy shit at fuck it ill just cheat somehow

what the fuck did I just read?

jesus

no, weird

:D

Thats the way i think, love it

Nice mindset of truth :)

'If someone can tell me how it should've been written. The message came across just fine.'

I'm going to remember this.

I've realised ESR is the only place where non native speakers fight to the death over who speaks English the best lmao

he meant "pubically"

"gamestyles were developing as a result of the netcode."

explain

I already did multiple times.
But long story short:
1) the higher the combined pings are the more rewarding +forward lg becomes (for both players). (the lower ping player can also play agressive with other weapons in general... but the other weapons somewhat lack the added bonus of the lg...)
2) The higher your own ping is the easier +forward lg becomes (there are some if's involved.. like get hit by a rocket and suddenly you're screwed... and a few other things)
3) the higher your own ping is the harder +back becomes. (especially backing away with lg = instant death backing away with rl can somewhat be done but is hard).

==> if any player has a fairly high ping the, by far, most favorable option to play is agressive with lg.


(rail is similar to lg but if you miss your rail vs an opponent then he will just jump in and +forward lg you....)

Nowadays playstyles (online) revolve almost exclusively on getting into a possition where you can easily agress your opponent with lg.

either i get more experienced, or you care more to explain your thoughts, cause i totally agree with this:

Nowadays playstyles (online) revolve almost exclusively on getting into a possition where you can easily agress your opponent with lg.

i assume its just one of the gameplay possibilities offline? looking on recent cypher performance at dh.

The only reason you would think that I didn't explain stuff is because people thought I was blaming illusions for "my own shortcommings as a player".

I explained this stuff fine and in a lot of detail....
But people were too busy flaming at me and trying to think up more and more insults about myself and about my skill.


Offline it's just a possibility yes, but it's harder on lan because the agression is harder to pull off and the defense is easier. Ofcourse there are still cases where it's the best option.
But cypher's skill is not so much based on pure agression... it's more that his overall speed is so high that he gains more ways to be agressive. Even some moves which I would find to be defensive seem to be agressive when he does them haha :D

I agree that LG style plays are definitely easier to pull off when you are on a higher ping, I've experienced that myself and tend to favour using LG over most other weapons if I have 50+ ping because I find myself hitting good accuracy on it consistantly despite the higher ping, whereas rail and RL are more difficult to handle.

Edit:

Somewhere in my or Weird's post history, he mentions about how 2 players, going around a pillar, that the one being chased can be hit and even die by the chaser's LG even though the guy being chased can not see the chaser.

I have not seen this but I guess others have. Your #3 made me think of that and if that's true, it would suck so much to lose like that.

there never was much development for the game itself in ql. it's still basically q3 with some changed values. I wish it was different but I gave up hope a long time ago.

being pleased isn't weird

Cypher rapes on any ping.
Personally, I am very satisfied but then again I haven't seen better.

cypher is basically neo though

and there's only 1 "The One"

neo is one spelled backwards

not sure if trolling or just stupid...


n e o

o e n



...how you figure?

old interweb joke

shux :/

>>> Cypher rapes on any ping.

Exception that proves the rule.

cypher aim sux tho
also there are plenty of games with better netcode
bf3
cs 1.6
firefall pvp
the newer cod games have a lot of interpolation and are pretty decent

cypher aim sux tho

hes the best aimer in this whole fucking game. deal with it noob.

lol tell him to beat tox in 1v1 lg
or even strenx
or spart1e
or heck even heartless who is out of practice a bit if i recall

he did beat them all. he is a quake live champion, unlike anyone of ppl you mentioned.

or do you want eye2eye lg battle, which has nothing common with real aiming skills?

pls hitsu jus becus u r permbad n blind at shooting(or fuck een duel) not mean every1 is
i've seen cypher rail
his reactions are not very fast

he relies so much on prediction for railing, which is why he gets rail raped by players like k1llsen in the open
he also hits 0 interesting rails

players like spart1e hit the most outrageous rails that you do not expect someone to hit, cypher hits 0 of those

he's just a rocket spammer

you think cypher could do any of these?
https://www.youtube.com/watch?v=Dbhrjy-iejY&hd=1&t=1m35s

this would be like spart1e's hometurf

ye, with ez. just matter of tweaking his sens.

ur just one of these guys who will wank to tox 50pc lg, and jump to conclusion cypher's aim is worse because he hits only 40.

but you seem to fail cypher does preshoot like stupid mother fucker, while tox doesnt.

the same goes for any player you can think off.

i don't give a shit about accuracay

hitting the angles and hitting the autofire is what i care about

cypher does neither

seeing some quality mousing is nice, spart1e is def the best out of the duelers

its like talking with someone whos blind.

jigglywiggly is right tho

i wouldn't say that cypher's aim sux, but still, compared to people like strenx tox stermy hal k1llsen or even spart1e and a lot of very good CA players, his raw aim isn't as good

cypher is like mario in the nintendo games, he's very good at everything, like he got a lot of experience, good brain, good positional, good awereness of the game, good combat skills (dodges & "situational" aim) excellent defensive rockets

but just like mario, he doesn't have any real strenght, I mean, he's not GODLIKE at anything, just good at everything. Except maybe his defensive rockets, which are very very good

so when you talk about raw aim, "pure" aiming, cypher is definitely not the best

actually cyphers strenght is being amazing with any weapon you give him

you can see you're a CA player in the comment....

You can play duel in a LOT of different ways. Cypher plays it in a way where he only has really small 'windows of oppertunity' which forces both him and his opponent to take difficult shots.

Strenx, tox, stermy, k1llsen (dunno about hal) base their game on getting easy shots on a predictably moving opponent who (preferibly) is in a fairly open space.

For instance on t7 stermy would often deal dmg when you pick up an item (as they are exposed) from a long range which makes his shot an easy rail shot. Or (especially toxjq) look for damage on the exit of teleporters.
This means that they can't move around a lot... or at least that your own movement influences their options.

Cypher plays in a different way which makes his shots harder too hit.... That he still hits a lot of them is proof that his aim is awesome.
You call it "situational" aim but it really is just aim.
If cypher would addopt the same playstyle as the others he would probably hit just as much.

k1llsen aim is very prediction heavy, his ca gameplay doesn't look anything like his duel gameplay so i will agree

tox does rely on predictible movement but not as much, but he also has an autofire for reactions, check his q4 gameplay with fatal1ty

if you want very raw duelers there's: spart1e, cooller, and ratpha
they shoot very often and go for hard shots

his raw aim has improved by a lot during the last half a year

did you try him @ eyetoeye duel or even thunderstruck ? :D

yes, he wasn't a really good lg aimer back then(ss kana/shootmania times) but now when you face him in tdm he hits really good, you can even check his accuracies a year ago and now, the difference is obvious

hal? u srs? he is can only hit a bit lg on t4 newbs, in duell he gekts rekt hard

LOLOL

he hits the shift key with his metacarp. (his hand. he hits it with his hand, not a finger. but a bone inside his hand)

r u 2 trolls really talking

warping players need to be "fixed" :/

this. they make me want to quit playing a lot :/

keep playing, just quit playing those players

Agreed, i hate wifi players so baaaaad.

lo...l u ma..d..?

https://www.youtube.com/watch?v=uFLdt_vkOGc

netcode totally shit. ping 50 vs ping 20 zero chance and no fun to play. its so random to hit.

for who? :D

when i have 50 ping i really feel the disadvantage against players with 20 ping.

Indeed, it's worse than it ever has been - which is hilarious considering that 50ms isn't that bad.

ive fucking 50 ping in my own country. that suxx. what the hell.

Well, at least you have servers in your own country :>

Hahaha :D

what if you have 90 ping and you are forced to play vs 30 pings ?

nobody is forcing you to play

what do i do then ? i stop playing ?

i could stand it some minutes

it's horrible
the rail thrus are lol

CPMA netcode I found was worse than QL, the lg did not reg well at all.

cpma netcode was influenced quite a lot by your config.
For instance xerpClients 1 + lagHax and xerpClients 0 without lagHax was similar to switching between osp and QL. (it's not equal but the difference should be as noticable)

try xerpclients 1

you should have used xerps clients 1 yo

All of the rails I hit do no damage. All of the rails I miss do damage.

???

this!

set cg_predictLocalRailShots "0" // A value of 0 will feel less responsive in high ping environments but may prevent wrongly predicted railshots and/or impacts.

I have it at 0

set it to 1. solved.

so that he sees shots hit that don't actually hit? pretty sure that's what he's seeing now, and that's what's so frustrating for him...

you don't understand the netcode m8....
see responce below on your other post.

Played a few games with it, felt no different.

really...?... you can't feel any difference at all?
That's amazing.

I thought I felt a difference at first, but as it went on longer it was all the same.

All "cg_predictLocalRailShots 0" does is 'delay' the rail beam animation by your ping, it delays showing the beam till the next shapshot is recieved IIRC. It's not actually helpful for gauging where shots actually land. In unlagged mode it'll actually lead to a far more inaccurate view of where players are in relation to shots.
If you consider how unlagged actually works, with regards to backwards reconciliation of shots, and interpolation, turning this off makes no sense.
You'll see where the rail was shot on the server at the point you fired, but it won't show you where the player was at that time in comparison to it. The way it's supposed to work is backwards reconciliation, it's supposed to check back in the server snapshots to see if the player was in the right place when you fired. So it doesn't matter where you actually SEE the rail beam. Same thing with truelightning, complete placebo in unlagged mode.
bla bla bla bla, you get the picture.

i actually don't get the picture...sorry if i'm being slow :/ When you say "it doesn't matter where you actually SEE the rail beam", i disagree - it would be much better to have what you see reflect what actually happened.

In my understanding, using cg_predictLocalrailshots "0" should provide a more cohesive audio/visual representation of the game, so that visuals match sounds (which both represent real events) i.e., when a rail is fired, you see/hear either: A) a railshot hit (screen shows rail hitting target, hitsound plays); or B) a railshot miss (screen shows rail missing target, no hitsound).

with cg_predictLocalrailshots "1", i think you're opening the door for seeing a railshot go directly on target, but because of ping delay, the shot actually missed and the visuals are incorrect.

looks like Marklar has a good explanation here, although i can't confirm it's correct, so please let me know if i'm believing BS.

"with cg_predictLocalrailshots "1", i think you're opening the door for seeing a railshot go directly on target, but because of ping delay, the shot actually missed and the visuals are incorrect."

What happens in ql (in non technical terms) is that the measured time in the underlined part gets "rewinded" by the server.
Aka the server 'remembers' the past 50ms (or so) and it then looks where the opponent was at the time that you pressed fire.
Therefor predictLocalrailshots 1 is more accurate than 0.

It's a horrible system though and I wish they would change it to something more sensible... but I've been saying that for years ;x

so you're saying i can control the way the server functions and interprets my inputs with this cvar? that is, i can get two different outcomes for exactly the same inputs (aim at spot x,y, press fire) with this variable set one way vs. the other? of course, those two different outcomes would be:
a) shot hits
b) shot doesn't hit

??

sigh... No the server does this regardless of any client cvar. It always looks at your ping and looks back.

Well thing is, what *actually happened* and what you see are completely different things, that's what I'm saying.
I'll have a go at explaining it.
The way 'unlagged mode' works, is by rewinding the server's logs by (yourPing)ms whenever you use rail, so that it can check if the enemy was in your sights when you pulled the trigger. ( since you actually see the enemy ~0.5*(yourPing)ms behind where the server sees him ).

That's why if you're playing someone with high ping who's railing you, you might get that effect where you get round the corner into cover and then get railed, seemingly through a wall. That's because when your enemy fired, you were out in the open on his screen, and the server checked and saw that he had line of sight with you when he pulled the trigger, even though that looks completely wrong on your screen.

Point I'm getting at is that if you have cg_predictLocalrailshots "1", it's drawing the rail beam's location based on the client's interpretation of where it should be, which shows where the rail is hitting on 'your time', as opposed to the server's. Theoretically, discounting any errors with unlagged, this is also showing the correct spatial relationship of rail to enemy more or less where the server would see them after performing 'unlagged mode's backwards reconciliation of the shot.

with this on you should see rails go more or less through the model to score a hit.
With it off, you should see large gaps between moving targets and railshots that hit. because even though it shows where the rail landed on the server's time. It'll be drawn (yourPing)ms after you pulled the trigger and the model will have continued moving all that time. And even though you were aiming at the model, and scored a hit because the server worked out where the enemy was at the correct time, the rail is still drawn late. It'll look more or less right with lowish ping in most cases, but in the long term, it's just visually confusing and unnecessary... plus rail feels heaps more responsive with cg_predictLocalrailshots "1" anyway.
It was more useful in a time where you had to lead shots because of your own ping.

Let me know if I explained that badly, or if you have any more questions.

i can play game with 60 ping. not too shabby

it's maybe cause you didn't try 35ms ping :p

i ping 30 on swedish servers

its not xerp but still fine.

playing vs lowping rockets heavy players can be annoying,

45-50 ping for me is the sweet spot for rockets tbh. :<

CPMA and WSW have much better netcode. QL feels like a mix between ra3 and cpma (with no xerp) IME, but as a player who can't ping below 72 to any European server I must admit its still playable and something you can learn to adjust too. TBH, I thought it wasn't a big deal at all until I played on LAN where I was surprised to discover my aim suddenly could hold up too other players who would just aim-rape me online, so there still is a bigger disadvantage then I originally thought would be.

I never understood why, considering the QL dev team hired arqon, they didn't just use CPMA as the building block for QL. It took about another 2 years and countless updates until we had all the competitive features which were available in CPMA added into QL. Though most importantly, until this day the netcode doesn't hold up, neither in terms of being as high-ping friendly while feeling consistent across different ping values and especially not when it comes to dealing with warping were the opponents of the player warping are hurt much more the player who is warping himself.

why? because id would have payed arqon and not sync ;)

would've been the right move.

and what kind of a programmer would name themselves sponge? Sponges are fucking stupid. They just sit there. I washed my balls with one last night.

and what about BOB ???

U FAIL

I agree that WSW netcode is superb, but QL netcode feels quite good as well. At least compared to OSP.

did arqOn really do all that much for Quake Live?
Didn't he come out recently in a post apologizing for being MIA and explaining that he had been dealing with illness when the QL dev team were trying to contact him?
IIRC, arqOn had nothing to do with the netcode or QL in general. I could be wrong, but I didn't think they ever hired him, or did anything with him.

I could be wrong, but I didn't think they ever hired him, or did anything with him.

arqOn really did some contract work for QL. I'm not sure exactly what it was but I think he touched the "netcode", yes.

source about illness?

I don't know how you can whine about 50 vs 20 pings - it feels almost the same for me - no problems with aim at all, you'll have some problems in 60-80 range - its casually playable for sure, but 50 vs 80 is already bad at competition. So its quite good, still I can't figure out - afaik Arqon made netcode for both QL and CPMA, but CPMA can be played with 110 ping and QL is unplayable there... wtf

oh fuck this again,

tons of lpb players talking about how everything is shit for them lol.

i love the netcode(only played OSP before, and not CPMA).

80ping is better than 20ping in ql

what about 10 ping?

What if sirax takes a look at it for a couple of hours then fixes everything, just like he did with all the qlranks related projects that id software never could implement.

hey guys it's sirax again. I spent a couple of hours o it and now we have a cure for cancer: www.qlranks.com/panacea

I don't know what sirax' specialities are but they can just ask me if they want it fixed... I mean I already told them several times how to fix it (even gave some lines of code and told them where to place it...)

so why do you think they didn't do it?

You should ask them....

I wrote quite a few comments on things that were broken which were heavily disputed by the developers while I was in the focus group.... And then got fixed (due to their change-log) after I had been kicked (not really kicked as in gtfo but due to inactivity as I was working on a few projects irl which took up too much time :p).

I kinda have the feeling that they hate my "I am right and you were wrong!" lines ;)

So you think they didn't accept your netcode fixes because of your attitude?

I hope that's not the reason hehe.

No I kinda mean to say that there was this whole "weird is seeing things" attitude among players and others (perhaps also at ID)... So my opinion was often sidelined in favor for others.

After I quit a lot of the things I complained about actually got worked on but I don't know if my whining about it had anything to do with it.

Usually these things go like this: I say what should happen -> everyone disagrees. -> half a year later someone famous sais the same thing -> everyone agrees -> it gets put in.

This also happened a lot while I was still in focus so I would point them to those things and say: "hey if only you listened in the first place... I would've saved you all a lot of trouble." <-- somehow I had this strange illusion that actually knowing and understanding everything about FPS would actually be a good thing for myself. (as I hoped to one day be involved in the developement of a competitive fps title.... but I really doubt that is ever going to happen though)

maybe they thought they know more about it then you do? maybe they are right? have you got an example of something you told them that got fixed later?

Sure last thing they fixed was the "cylindrical hitbox".

They always said that it was a cylinder.... I told them, on irc, (even before they rolled it out to everyone) that there was something wrong with it. I think I said something like "it feels like a capsule rather than a cylider as the midair rocket explosions are messed up in a way that makes me think hitboxes are capsule formed"... They said no... They even went as far as to enable drawing of the hitbox around the model (only for testing on their 'test servers', which btw was quite informative) AND nowadays it's fixed. (like more than 2 years later?)

But a lot of things got fixed... I remember playing several maps with a shitload of item layouts while I had already said that they should look like what they now looks like.

But I kinda feel uncomfortable sharing this stuff as I don't know how much I am allowed to say about my testing days.... And I don't want to discredit any people (which I will if I continue this haha :D).
Could we not discuss the internal testing stuff further? :)

you are truly the cat's meow

Your signal goes 1000 miles over copper and optical wires underground and underwater so that you could rail a green person.

And It works !

Let's all take a second to appreciate that

Internet - my friend ; _ ;7

AND IT TAKES ONLY 50 MILLISECONDS DELAY

It's shit. Enemy's ping shouldn't alter your own hitscan behaviour but it does a lot.
Makes you not wanna play the game serious when you have the perfect lock on and yet you only hit 35 instead of 50+.
50+ms, autonudge and maxpacket 63 and you are good to go ...

Noobs who hit 30 shaft with tjernobyl internet Will dodge(warp) like cooller vs "lanpingers". And when they really play on lan they get crushed.. Been like that for years .. Ruined a lot of games for me when i was playing.

hi Weird

I don't post anonymously.
Nor do I fakenick. As imho both are a kind of deciet.

But anonymous here is fairly correct though. I wouldn't say that it "ruined a lot of games for me" I just got really upset about the games which were influenced by this. But "in my time" (rofl) that wasn't too many games yet.

I could just tell that if I wanted to win more I had to play a certain style which I disaproved of as I couldn't see it working on lan.

S'up weird

the higher my ping the more easy it is to rail...

cg_projectilenudge 1 with cl_timenudge -12


= pure bliss

projectilenudge 1 does nothing basically.

try 40 or 80

projectilenudge does nothing unless cheats are enabled...

oh, didn't know that :o

but when you set it to let's say 40-60, it doesn't say in the console "this cvar is cheat protected" or anything like this.

weird.

very weird...i was wrong - first time i tried in a regular/ranked server, is said "unknown variable cg_projectileNudge", but i just tried again and it was fine. my bad :/

It's shit hate it, cpma netcode was by far the best.

I meant -1

apart from model and hitbox sometimes waaay off sync(rail), but mostly just a bit behind, i guess it works. The pillar statement is still very real because of this though, happens all the time if being chased closely. Yes, say use rockets, but often they just go through +forward players

Timenudge doesn't work in standalone. That was the thing which kept the game playable at high ping.

60-80 ping and with timenudge -20 you could still get pretty accurate position of hitbox compared to the model position, even if it was moving very fast.

If a player was warping, then timenudge alleviated aiming as well. At least you'd get his accurate position between the warps.

Now it's all lottery, very frustrating.

Barely a point playing the game for me anymore personally and I've been playing since open beta.

Timenudge doesn't work in standalone.

and where did you get this amusing idea?

net_restart doesn't seem to work neither :(

before standalone when i had 55ping on DEU i did net_restart a couple of time and then bam my ping dropped to 47ms :O

But are you sure about timenudge tho ? Cause when i set autotimenudge to 1 it adds some yellow lines in my lagometer

net_restart is an interesting case. Majincline found a very good explanation. Basically, net_restart is supposed to close the socket and reopen it. The normal behaviour is to use the same port and this is exactly what QL does right now.

When there's a problem opening a socket, the port is incremented and it tries again. On the plugin version, it seems like QL often failed to open the same port again. It probably wasn't properly closed by the time it tried opening it again... some weird timing issue... we'll probably never know why exactly.

Summary :
Plugin QL : net_restart uses a different port.
Standalone QL : net_restart uses the same port.

Now, why is that important? Apparently, using a different port had a very funny side effect. We suspect that some node in the way your packets take have port-based load balancing.

So, anyway, to reproduce the behaviour of plugin QL's broken net_restart, Majincline suggests using something like this :

bind ' "cvarAdd net_port 1; net_restart"

Incrementing the port seemed to replicate the old functionality -- A small hiccup in the netgraph, followed by a noticeable difference in ping.

Since actually measuring such a thing is rather wishy-washy, I'd be curious whether anyone else notices results.

If it does work, I'd recommend a set net_port 27970 in autoexec.cfg to keep the port from getting too big.

thks man

cvarAdd net_port 1; net_restart

actually worked \o/

i get the red vertical line in my lagometer and i lose ~4-5ms :)

so hard to find usefull post right here
ty for giving ping 45 insteed of 57 back to me :D

That's 12 ms of difference man, that's huge, I can't make this work :'(

It doesn't work for me :(

I hit the key I made the with, but then nothing happens, does it mean it's working only for some people? Am I doing something wrong? :(

HALP

What's the exact behaviour when you press? Are you sure you didn't mess up the bind? You can do it directly from the console too if you want.

If it really doesn't work... I'm afraid I can't help you.

When I press the key:

"net_port will be changed to xxxxx.xxx upon restarting. Opening IP socket: localhost:xxxx"

And I check my ping, nothing is different. But it's ok, I'm used to that ping anyway :p.

Well then it seems it's replicating net_restart behaviour from the plugin era. It doesn't guarantee that changing the port will give you a better routing, though. If there's load balancing based on that, it may not be decided with even/odd numbers either. You can always try experimenting with many different ports and see if you find a scheme (or change your routing at all).

Holy coooooww it's actually working, fucking thank you FlashSoul ROFL xD.

[+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+][+]

Edit: btw will I have to do that all the time I join a server then?

Edit: btw will I have to do that all the time I join a server then?

There are too many variables to tell at which rate you'll have to do it. It depends on if your net_port stays the same or resets itself to some other value. It depends on if you are connecting to the same server (or passing through the same node with the load balancing). It depends on if the network and the values on the nodes on the way to route you changed. It depends on a shitload of things, really.

In any case, if you worsen your ping while net_restarting, you can always come back to the old port. You could even have cvaradd 1 bound on + and cvaradd -1 bound on - or something.

Good, thanks for taking the time to explain! I will look into adding +1 or -1 later :3.

No problem. It's always a pleasure to help.

longshot but any chance of there being an equivilent in source engine games!?

I have no idea. There are probably TF2 geeks somewhere that would know.

cl_timenudge still works in the stand-alone, but he's right about cl_AUTOtimenudge not working with the new stand-alone and might have confused between the two.

I must say, heartless does know how to get an ESR thread going.

The game feels like shit on every server. I get the best performance on texas servers where i ping 70-90 lol.
It feels sluggish and severely effects my movement. It's like trying to run in syrup and sometimes it feels like you slide on ice.

2009-2011 i might encounter 3 warpers in a week... no big deal... now it's 5 out of 8 players on a ffa server that warp....every game.... The game has become unplayable.

I don't get any frame drops... i can run any map at 250 fps without hiccups... yet it feels like i'm getting 25 fps.
It's literally the worst online gaming i've ever experienced!
The game feels/plays like utter shit.

I actually thought that moving to a standalone would fix a lot of the problems. It didn't. things are far worse.

I thought they turned the game back into an .exe...and got rid of the plugin bs. they didn't. all they did was make theyre own browser and wrap ql in it. So it's virtually no different than before.

The only thing i can compare the way this games shittyness "feels" is to compare it to when my clanmate used to run servers in q3 on his home pc back in 2005-2007. His connection was roughly 400KB Down/80-100KB up. If there were four or more people on the server and everyone had their rate set to highest... the "warping" would start... So we would all lower rate to 3000 and it would solve the problem.
As far as i know they haven't changed the netcode at all. They did change the hitboxes a few years back but that didn't cause and warping.

So comparing from past experiences on my friends server to how ql feels... like a bandwidth issue not really netcode. The game used to feel smooth and awesome.... so what changed?

I expect more from an AAA game company. To bad half the company abandoned ship... even the captain!

"It's like trying to run in syrup and sometimes it feels like you slide on ice."

sounds like "cg_nopredict 1", but i doubt it.

I think server hosting quality, and in genral the whole isp grouping are contributing quite a lot to the mess. Server in frankfurt has been terribad this last year.

Hello fellow swiss person. I'm getting 120 ping to GER and NL, 48 to Milan (R.I.P.), 55-60 to GBR, about 90 to UKR and 80 to SWE. How about you?

Do you live in Bern, have adsl? My office mate does and has similar pings.

I live in Zurich and recently switched to optic fiber. I used to ping 8 to Milan, and ping 15-30 to Frankfurt, 30 to GBR, 50 to madrid and warsaw.

I agree, especially when there are a lot of people on the server. It'd be very interesting to see how the game would play out on simple 100fps servers (not to say 1000fps, which has been the competitive standard in cs for ages) and whether the issues with the netcode would diminish or even dissolve. I really don't understand why they don't offer such an option for Pro subscribers, clearly this would be a real added value (even if the price for it had to be increased).

QL has a bit better netcode than q3 osp and nothing more. All other games/modes like qw/cpma/wsw have more than 1,5 developers (that QL have) that's one of main reason why many things a still not fixed.

QL's ping compensation for instahit weapons (mg/lg/rg/sg) work well with ping <~80ms and with timenudge -20. For example at modern qw (with antilag) there is not limit (or this limit is too high) for instahit weapons. I was surprised when i lose lg only matches versus americans or brazilians at qw's povdmm4.
afaik projectile weapons cant be fully ping compensated like instahit or they can but this will lead to fully undodgeable projectiles, sudden teleporting due knockback and other shit while playing against players with much higher ping.

I'm personally semi-satisfed. I like the fact that it's not like baseq3 or OSP/RA3 where low ping players had a serious advantage even over mid ping players.

ofc. At q3 osp there was not any ping compensation from default u have to hit rails in predictions.

Best ping range for hitscans (LG/MG) is probably between 35 and 45ms. With such a ping, it's basically easier to hit the opponent if the crosshair is right on him.

bullshit. after i saw how heavily aimed player hit same 40-45% lg with ping 80ms i had no doubt that ping compensation for instahit works well even at 80ms. he was using timenudge -20 like many top players do it any ping.

For rail, well, it depends, sometimes you have your crosshair right on him but it doesn't hit still, and sometimes your crosshair is quite close to the hitbox and it's still a hit, and that no matter the ping range you play with.

Yeah ping doesn't matter. From my experience i hit much more rails when crosshair not on enemy (but too close to enemy) than i didnt hit rail when enemy was on crosshair.

Timenudge and projectiles nudge can help in this case

Timenudge works even at 5ms ping. projectile nudge doesnt work ihmo.

uber9petska: btw cg_projectileNudge and cg_autoProjectileNudge work?
@sponge: dont think so, dunno
@sponge: i dont mess w/ that
Drayan: they were disabled serverside some time ago
Drayan: or I assume serverside
Drayan: they don't work either way
Drayan: it's pretty obvious when they do

from ql focus irc chan.

Pls, id, and especially SyncError, could you tell us if there are plans to modify the current quakelive's netcode ? A netcode like the one you can find in CPMA would be PERFECT.
CPMA's netcode is so perfect in many ways, and with such a netcode, some euros players could even play with the Americans because it would still be playable with a ping =~110ms.

Will be good if they will make their server's ignore clients timenudge when ping below 5ms. No need judge/admin to ask players to set cl_timenudge to 0 at LAN. That was At Prague TDM LAN :\

I'd just like to point out that high-ping players are currently teleported by low-ping players' projectiles (and hit-scan weapons with pushback!), so it would only be fair if low-ping players were teleported by high-ping players' projectiles too!

I think the net-code is fine. It's an internet game, and as such you should expect odd and unfair behaviour. Players should change their style of play to suite the environment. If you know you can't use rockets/shaft/rail very well on this ping don't position yourself for rockets/shaft/rail use. Be more passive if you have to, or be more aggressive if you have to.

People get beat, and they blame the net-code. They write an essay on how terrible the net-code is, and how it was the net-code's fault for their untimely death, etc. It's all pointless. We can't change the net-code with these long whines and rants, but we can change how we play. So focus on that. If the developers improve the net-code then that's great, but if they don't it's still O.K.! It's still a fun, challenging game.

yeah high ping players suffer from projectiles now. imho it's silly to make same for low ping players (to suffer from projectiles of high ping players) this will make game much more random. it sounds like making 60hz cap to stop 60hz user suffer from players with 120hz monitors. yeah this will make game be more technically fair(equal) but this will lead to a big decrease in players...
i think netcode is fine too but can be improved for sure. it it's not a rocket science.

imho main problem: QL has 1,5 developers and they can't or don't want use help from community.

Quakelive does severely need a netcode overhaul. Australia + New Zealand have some of the worst internet infrastructures in the developed world, and Quake Live's netcode is just not adapted to dealing with any adverse conditions. Some players are nigh on unhittable / undamageable. The issue of warpers has been getting progressively worse for some time. So much so that many players from specific regions are absolutely unhittable... That could also just be due to bad routing or servers... but as I said, the netcode deals poorly with any adverse conditions.
I could speak volumes on it really...

In comparison to other game's like CS/TF2 I personally think its probably the best out there. Just from my experience I found that CPMA/Warsow feel similar to how hl/hl2 games feel where its fine with low/equal ping but once you hit above 70/80+ sure it plays better but you get into strange territory where high pingers actually can get some form of advantage eg. They can slightly skip/move faster briefly(regardless of having client prediction on or off on your side) yet still can hit very accurately.

For w.e reason I have so many memories(nightmares?) of random 200+ pinger's in CS doing a crazy fast peek killing a team mate and disappearing or magically skipping past my pre set crosshair throwing off easy timed shots.

I think modern delag features put into OSP's built in latency netcode would get rid of most of the random rail things happening. With a fixed latency, delag would be attempting to get everyone to have a 50 ms feel which is more plausible than trying to give every one a 0 ms feel. Low ping was a huge advantage in OSP but it didn't have any thing but timenudge to help high ping people, and even with timenudge it was extremely stable none of the random things you see in Quake Live happened in OSP. Maybe weird can answer if this is true or not.

I think the whole point of any delag is to bring a gap of low ping and high ping together well if you make that gap smaller by putting in fixed lag wouldn't it lead to less randomness?

If you look at the worst delag systems its the ones who try to correct everyone to a 0 ms, 30 ms being corrected to 0, 80 ms being corrected to 0, it would basically be like both players seeing different games on their screens its so far apart, with fixed latency the correction point would be 50 ms, 30 to 50, 80 to 50, much smaller corrections on both ends.

You're a bit misinformed about osp but if you were to force players to play with a small default delay, then yes a lot of these issues can be negatted.
You can pretty much say that if you fix shots delay to X ms that you can make all pings < X very similar to play on.
Q2 is a fair example of how delayed shots can already equal a more fair playingflied without any type of netcode fixes. If you were to do some smart calculations you can make it even more fair.

Is that what you suggested to id or something else?

It was not my first suggestion to them... I have mentioned the possibility of it at some point in time though (in some discussion on irc where I think I named it "a funny alternative").

In general I think that such a solution would influence the game a bit too much in favor for the defensive player... At least if the conditions are "lan like" and you still enforce this it would influence the game too much imho.
(I am not in favor of actually changing the lan-balance of ql... it plays godlike on lan and I wish that I could've played on that more often as it really is very different. Also I'm not in favor of having a different setup/fundamental game online than on lan.)


My personal preference would be that each computer has a cvar "like autotimenudge" but with which you can set a scaling aka autotimenudge 100 = 100% and 0 = 0%. Where if you take 100% your pc will extrapolate the opponents possition such that if you hit him on your screen he will (given that he follows the same trajectory as predicted) be hit on the server. If your pc predicts wrong though (which will happen on high pings quite frequently) the opponent will seem to warp a bit.
But in this mode you must disable the anti-lag / backwards reconciliation as otherwise people have 0 incentive to pick this option.
That way people can setup their own computer in a way such that they find the warping acceptable while not having to lead their shots by too much.

For people who have a high ping or warpy connection that would mean that they either accept a bit more warpy opponents or aim ahaid of their opponent a bit.


But in general it doesn't matter what I would find the most acceptable solution. Any solution which solves the problem in a sane way will have my approval.

I just can't understand a solution where a player with crap internet is punished less for this than his opponent who has good internet. <-- this heavily favors everyone fucking up their connection... and trust me I have played tons of officials vs people who put torrents on to get further in the tournament.

Also I think that the choice of netcode should be consistent with other things.
Which means that IF they decide to stay with this netcode --> they should implement predicted hitbeeps.
They should also remove timenudge or make it such that it only works when your ping > max ping for which backwards reconciliation works,

(
edit addition: Also things like cg_projectileNudge are really sane things to have... that ID disabled that (at least according to sponge) is just mindblowingly stupid. They can still alter the implementation as atm it just shifts the trajectory (which makes the projectile have a huge warp when it gets fired from the weapon and then after that be accurately draw) into something more gradual (for instance during the 1st second the speed of the projectile is slightly higher such that at the end of the second the ping is corrected ( so say (estimated) you ping 50ms = 0.050s = 1.05* regular speed for a second)
)

That's good and if it's done right can make more equal diff pings.

that's because q2 is the shit!

server are fucking craps ( UK) lately and there are less and less prem servers available :/

l4d3 confirmed

the worst issue is med-high ping players warp but their low ping enemy is fluently for him.

seta cg_NewNetcodeDeveloper "Weird"

I think the programmers at ID are qualified enough to implement this stuff... I am not like some monsterly great programmer (I wouldn't say that I'm bad at it though ;p).

I just think that they don't believe it to have the impact that I've always said that it would have...
Or maybe they disagree upon some level with my logic but at least I haven't heard a sane argument in favor of this netcode which I haven't refuted by pointing to their implementation combined with ingame situations etc. ;)

How high ping friendly would the solution you propose be?

Personally I don't think that the situation is that higher ping actually allows you to hit more lg, it just hinders all other weapons and thus doesnt allow you to pull off attacks/defensive plays with projectiles/etc that you could on lan, but you could still deal just as much damage with LG on lan.

Btw, if you set up a custom server, I'm pretty sure you can tinker with the g_lagahx values (default being 80) and be setting it to 0 perhaps replicate osp style netcode feel with no anti-lag.

well that depends a lot on the player.

I think that up to a ping of around 30 you won't be able to see the warps...After that you'll get situations where when the opponent strafes left/right that he will appear to warp slightly... The question is kind of how badly can you take warp vs how much infront you need to aim.

For instance aiming 20ms infront of somebody is not a hard thing to do (in most cases it's really easy). So you'd already be close to perfect for any ping up to ping 50.

After that it's all just preference.... I personally prefer to predict players myself rather than let my pc do it for me.

But like I said I'm fine with any implementation of the netcode as long as it's done in a sane way.

would you say you are like top 4 programmer ;p?

btw.... I find it strange that I always get flamed at when I bring up netcode... and that if someone else starts about it I get a ton of +'es....

Something seems off there for some reason :s

I'll start another netcode thread for 10 easy daily payments of 2 [+].

sweet ;)

i hate how LG/RG are unlagged while shotgun is based on your ping. it makes 0 sense

but overall the game experience is too closely related to ping, and it's one reason why the game is dead (in NA)

I'd rather see 100 ms minimum latency for all players so that the experience is consistent whether you play at a close server or a slightly farther one. the community fragmentation is a disaster.

All hitscan weapons are unlagged. The shotgun animation is just delayed as it waits for server confirmation of shot. The actual hitscans themselves are unlagged AFAIK.

timenudge is horseshit.

/thread

The only thing I dislike is that timenudge makes the game inconsistent.

Don't come with a technical explanation, because all players know it does and that it was forbidden from lan play for a reason.

Ehmmmmm..... if you insist then I won't give a technical explanation but the main reason why it was forbidden on lan is because there was a bug in q3 where packets would be ignored by the server when players ping < timenudge.

(I don't know if this bug is still in ql... I think it's fixed... Also my above example is not extremely correct because then I'd have to become technical but it kinda paints the picture.... There would pretty much only be 1 update on the server for each snapshot if you used tn-20 on 0 ping... Sometimes there would be 2 and sometimes there would be 0... but both are considerably less than the 3 which are default. This is very probably the cause of the "blue bug" (named after the swedish player blue).)

See that I almost gave a technical explanation.... damn.... ;P

I wish this technical explanation plz, I want to know how the TN in lan ping!

And what is the history of this "blue bug"?

the history of the "blue bug" is that he played a match @ some lan on dm6 and t4 where he insanely clearly shot the opponent down but the opponent took 0 damage.
He ended up losing the game and it started a lot of controversy about what might've caused such a thing.

As for the technical details... I forgot where it was in the code...
But basically the packets a client send have a timestamp on there which is depending on the localtime and the timenudge.
(something like cl.serverTime = cls.realTime + some delta - tn).
This is put into a packet through something like cmd->serverTime = cl.serverTime.
Which is then sent to the server... and either the server at one time checked if the server time (svs.servertime or sv_serverTime or something like that) was smaller than the cmd->serverTime.
Or the client itself had some check like that in there which forced the cl.serverTime to remain on itself which then got ignored elsewhere....
I don't really remember... it's been several years since I checked it.
I'm not gonna walk through the entire code again to see what I meant in jan. 2014 ;P

I don't know the history with Q3, but iirc QL had tn on lan for a few years, then it was locked. As I understand it, some pros complained because they (like most) weren't using tn on lan, but some tryhards were using tn -20 to gain an unfair advantage. It's not really hard to imagine why one person having 20ms of prediction on a lan while the other doesn't would be a problem.

Something like tn shouldn't be user adjustable at all. If the game is going to use that type of prediction/extrapolation, there should be some prescribed basis for this and everyone should play by the same rules. The 20 ping trolls using tn -20 to face shotgun and face rocket people around corners, and perform invisible peek rails, are extremely aggravating. The ones that do so on wi-fi deserve a permaban.

They did introduce cl_autotimenudge a good long while ago now, but didn't force it on, or remove the user configurable version either :/

Tbh I'm glad they didn't force it on, it really highlights other user's lag, even visibly showing micro lag clearly as jittery movement.

there are so many misconceptions about timenudge and unlagged stuff it's insane...

To be absolutely clear I meant if you yourself use timenudge, poor quality connection of other players becomes easier to see.

They worked on making it smooth but if an opponents connection isn't just slow but jittery it doesn't cope well.

All jittery connections make stuff unplayable.
And yes timenudge forces an extrapolation and thus the jitter will seem to get worse. (with any type of extrapolation all kinds of jitter stuff will get worse, as the jitters are extrapolated/amplified)
But with unlagged their jitter is countered by the unlagged part, which is why nowadays everyone warps.
Throw out the unlagged (partially or whole) and people will stop playing on wifi as their shots will become irregular compared to their own view.

Maybe it's time for a little recap how the netcode works.
Please correct me when I'm wrong, I'm summarizing my observations here.

Game servers process gamestates (aka snaps) at a fixed rate of sv_fps, which is atm 40 (states/second) and will be configurable going forward. 40/sec means the time between two snaps is 25ms.

Now, the QL client doesn't immediately move everything to whatever coordinates it received last. Instead, it uses the 2 last received snapshots and smoothly moves everything from A to B for the next 25ms. In fact, this adds 25ms latency to what you see on your screen. This works well as long as packets come in regularly in about 25ms intervals. Whenever a packet is delayed (due to jitter or whatever), QL needs to extrapolate from A over B to an estimated point C. When the snap finally arrives at your client, everything is moved from the estimated C to the actual C, which is the warp you may observe.

What negative timenudge does, is to cut off a piece of the animated/interpolated 25m from A to B and move everything closer to B immediately and extrapolate more towards C. The effect is reduced latency at the cost of larger prediction errors and larger warps.

So with negative timenudge you will be able to see opponents earlier, but you might see them at a wrong location because it was extrapolated. You think you hit a player, but in fact, he wasn't even there.

Initially timenudge was intended for dial-up connections and use positive values, which will buffer packets to reduce the effect of jitter, which has the effect of adding x to your ping.



With unlagged I assume you refer to backard reconciliation, where the server keeps a history of gamestates and when it receives a packet from a client that he has fired a hitscan weapon, the server pulls out a historical gamestate based on the packet's timestamp and checks whether that was a hit.
This part of the netcode allows people with high ping to hit stuff, but causes you to get hit after you just escaped around a corner. That's because it took 50ms from the attacker to the server (if he has ping 100), up to 24ms for the server to process it and then 30ms (if you have ping 60) for your game client to be notified that you have been shot.


QL uses unlagged only for hitscan. Other games also use it for projectiles, which is a terrible idea imo. That would mean that when someone with ping 200 fires a rocket, you wouldn't see it for 100+12+30ms. First thing you'd know about it is that you're dead.
Try out a UE3 game, high-ping players with rockets are a nightmare.

IMO unlagged is only a problem when someone is able to receive packets without jitter/lag, but sends out packets with jitter/lag. Then others won't be able to see him properly (he's warping) but he's able to hit stuff without problem.


Appreciate any feedback/insights that can help improve my understanding of the netcode.

you've got the main idea with the exception of the already extrapolated own state of the player.

By which I mean that whatever action you do (in all quake series) have an instant effect on your own movement at your own pc. So in this sense your local location is leading the server.
In almost all cases this "lead" is perfect (as it contains what you yourself will send to the server). (only case where it's not is when you have some knockback due to getting hit)

This heavily increases the time between the points in time where "player A sees player B" and where "player B sees player A". Depending on the situation.

For instance if player A has ping 0 and player B has ping 80 and B runs into a room where player A is at. Then player B will really see player A at least 40ms earlier than visa versa. (there are some minor exceptions here due to reduction of data sent by the server but let's not bother with that).
If player A has a ping > 0 then this will just add to the 40ms.
Therefor you can have the situation where player A and B run around a pillar where A can never see B but B can see A (where B is the "chasing/attacking" player).

This leads to the insanely stupid gameplay where both players try to overly rush eachother with LG..... Which made sure that LG heavy idiots win all online cups.

p.s. negative timenudge reduces this effect (your local "game-time" gets pulled towards the "game-time" that the server has).

thx. The extrapolated own state certainly is a key aspect that I wasn't aware of.

As to the effect of timenudge in this situation:

With tn=0 it would actually take 40ms (pingB/2) + 12.5ms (avg. delay before server processes input @ sv_fps 40) + 0ms (pingA/2) + 4ms (avg. delay before client A processes snap @ 125fps) + 25ms (timespan between 2 snaps @ sv_fps 40 used for interpolation) = 81.5ms for a still standing A to see an incoming B.
With tn=-20 it would reduce the interpolation delay from 25ms to 5ms, allowing A to see B 61.5ms after he enters the room.

Is that correct?

hmmmm... there are some notes there ;p

For instance the server "processes" the input asap. That is when a packet arrives the server always processes it directly. (at least I currently think so... but it's been a while since I looked at the code... I might be not remembering all details exactly :p).
So the server state is continuously updated.
The server state is only communicated however once every sv_fps. (which does lead to an average 12.5ms delay)

The 25ms timespan between 2 snaps is also related to the 12.5ms... If you send a packet to the server which arrives just 1ms before the snapshot is send then this will be the last state of that snapshot, which means that it will be the final state. At which point the receiving client will have to wait 25ms to see it (after receiving it).
If the server receives the packet 1ms after it send out a snapshot then it will be the starting state of the next snapshot and it'll only take 1ms before you get there after receiving it.
So in general it's easier to just think about the whole "snapshot" stuff as being 25ms total.

So I'd say it's more around 70ms...

With tn-20 it would be 20ms less yes. But with the note that if it's "at the start of the snapshot" that it heavily depends on how the previous snapshot was extrapolated.

Anyways a while ago I made this post http://www.esreality.com/?a=longpost&id=2313076&page=1
Maybe worth your time to look at that.... back then I did have (most) details checked.

Using any form of nudge has pros and cons, and influences playstyles. On a LAN it would be an unfair advantage, considering the connections are perfectly clean. I don't think it makes sense to have a game where people are playing by slightly different rules. Weird has claimed nudge can be good if everyone is using it, but I haven't tested that. It definitely has strange effects with some laggers/warpers. Reflex is proof a combination of extrapolation and unlagged can work really well, but that might not be possible on QL. In that case only allow nudge above 80 ping and get rid of it from the game otherwise.

Play an online tournament or two and you will rethink your opinion.

Which opinion? I mentioned a few things. Plz elaborate....

"Something like tn shouldn't be user adjustable at all."
+
"...only allow nudge above 80 ping and get rid of it from the game otherwise"

You aren't making an argument for why it needs to be user adjustable. My argument is: anything related to lag compensation should be done on an automatic, logically prescribed basis. I also said I'd be open to a combination of extrapolation (nudge) and unlagged always being on (Weird has suggested this), but if the result was too controversial, then an alternative solution would be to limit nudge to high pings. As it stands now, it's being abused by some people and resulting in some idiotic playstyles, namely by low pingers with tn-20.

people should just always favor what I suggest ;)

i love cpm but it's netcode is too forgiving for vq3

I think its more so the old square hit-boxes which are just too easy to hit when combined with the anti-lagged netcode. The same problem existed in QL at first too which is why the hit-boxes was changed and reduced in size.

All i know is that having 8 people hit 48-50% mg on q3dm7 in cpma was ridiculous. You're right though, it was similar in mid 2009 for Quake Live too with the old hitboxes.

Good'ol times.

copy paste warsow's netcode, tell no one.

But really i'd be interested in seeing what ql runs like on a dedicated physical core of a good server at a good datacenter. I think a lot of the problems have to do with the game running on cheap virtual machines at 40hz. Retroactively changing maxpackets to 125 on people with old configs would help too since the change to make that default was recent. The game definitely could use a better interpolation adjustment method and have it on by default because autotimenudge sucks.

The servers run on dedicated (non-shared) hardware and not on VMs.

I really doubt each server has its own core though with the pro server spawning system though.

Well of course. That would be massively overkill and those cores would do nothing but sleeping most of the time.

I stopped reading when I read more than 5 mistakes. Got about halfway through the post though.

care to point out what i'm wrong about other than the virtual machine part? you've agreed with me in the past on making timenudge adjusted automatically.

Well the obvious ones are
1 all the servers are fine and 2 generally located in good datacenters.
3 The game runs (almost) dedicated (not entirely sure if they do something with it actually).
4 Q3 ran on 30 hz for a really long time (only got upped to 40hz on some servers) so that's no excuse either.

The really extremely "wtf is that" is "The game definitely could use a better interpolation adjustment method" because interpolation = calculations done between 2 known points... Which contains extremely minimal room for errors already.
You want a better extrapolation method.

Next to that I would be inclined to say that autotimenudge actually works fine. The reason why it doesn't work well online is because it gives you the disadvantage which your opponent now has. So basically you get 2x the netcode crap while your opponent gets 0x the netcode crap. If they however remove the unlagged then both will always be on 1x netcode crap (but based on your own ping) = a much better overall feel.

I never played q3 apart from cpma so idk how its netcode is but I heard it's even worse. I can tell you plenty of other games have better networking than ql though. ql is the only fps I've played where it is totally impossible for US players to play europeans because the game just falls apart at higher pings. The servers aren't always fine. If you have one spawned server with a bunch of people spamming weapons at max fire rate it ruins the other servers at the same location. The datacenters certainly are not good in the US at least. There has always been problems with players getting much high pings than they should be for the location due to poor routing service, although things have improved over the past few years. There's also evidence of poor maintenance like the 3rd person spectator model being on servers even though it got patched out a year ago. 40hz is a joke these days too. csgo runs at 128hz. The fact that people pay a subscription fee but can't get better servers than even the free official ones in $3 fps games is sad.

When I spoke of better interpolation methods I meant automatic adjustment of packet buffer size. Autotimenudge always makes players warpy for me, while time nudge in general does too but not as badly. It could probably use better smoothing too.

Idk what you're refering to when you say netcode crap, and I don't have intimate knowledge of ql's netcode but my assumtion is timenudge is a modifier for how far in the past the client runs.

" There's also evidence of poor maintenance like the 3rd person spectator model being on servers even though it got patched out a year ago." <-- what the hell does that mean?

"Autotimenudge always makes players warpy for me, while time nudge in general does too but not as badly. It could probably use better smoothing too." depends on how you set your regular timenudge (and possibly on what your ping is, although I'd have to look at the implementation to be 100% sure about this... I'm fairly sure about it now though ;p).
Autotimenudge and timenudge work in exactly the same way though (although 1 might be capped while the other is not).

Your assumption on what timenudge does is right.
But you can find all of that stuff online m8... or in my netcode writeup...

QL falls apart on high pings for, in general, other reasons than what you're claiming though....

I was talking about it falling apart being a problem in the general sense, not due to any of those issues. The hit registration at lower pings isn't that great either but is made up for because the hitboxes are simple.

If you go into freelook on a server and try to run into a wall, there's a decent chance it will stop before your camera hits it because it's in third person mode. This was patched out a year ago but still occurs, so I'm guessing they have issues even keeping the servers updated.

Autotimenudge is capped same as tn (-20) as of some recent build. maybe standalone was the first idk.

Care to elaborate on interpolation? My layman understanding comes mainly from another game (hl2 engine/cs source) and is based soley on experience.

In the early beginning of cs:s you had the option to turn on/off interpolation on your client (I think cl_interpolation 1|0 did the job). Usually servers were running with 33hz (called tick rate, but I assume both terms correspond). The game had many similar issues to quake (shooting through the model/hitboxes behind the model). Now if you turned cl_interpolation to 0 you could very clearly see how the models were lagging, because the raw server data was displayed without client treatment. If one could bear the lagging, then, in terms of precision, this actually improved the playing experience. Soon people upped the hz on their servers to 100, which in turn drastically improved the whole experience. The game became extremly precise and models ceased lagging on cl_interpolation 0. From your understanding now, would upping the hz in ql have a similar effect or do you think the effects are negible given the way the netcode works?

Just google interpolation... it's an easy to understand concept and it's well known etc.
Basically you take 2 points in space and time (aka x = 4 @ time 1 and x = 8 @ time 5) which are based on what you recieve and when you recieve it. And you draw states as each individual stepsize according to how fast you can draw.
aka if your pc can draw 1 image in 1 time and you recieve the above then you'll see
x = 4 @ time 1
x = 5 @ time 2
x = 6 @ time 3
x = 7 @ time 4
x = 8 @ time 5

Ofcourse more sv_fps are better... but in general q3 used to do a good job at solving this fairly nicely already so I take it ql does as well.

Posting in a revived thread...

they haven't used vm's for hosting for many years, even long before your post was made in 2014, as those servers ran really badly.

Maxpackets was forced to 125 since July 9th 2014 so you did eventually get that wish at least :)

I think from this topic we can conclude that not too many players are all that happy with the netcode and the way the servers perform/deal with adverse conditions.
Didn't Rapha even say that the hit detection on LAN worked poorly?

that was his excuse for losing that tourney lol.
He did make a few comments though.. through rails ect.. missed rj's

Typical rapha, making excuses..

[/sarcasm]

the hitcode on lan is also broken

there is some clip where spart1e gets a thru rail on evil and spart1e starts screaming and rapha saw it

rapha started to feel sorry for spart1e

http://www.youtube.com/watch?v=zXbRrTu6jP0&t=33m53s

It's like some drunk fuck ate quake3... shit it out... then when they woke up... molded said pile of shit into quake live... and forgot to pick the corn out.

90-100 ping > 30 ping - see : sn1ke

uh, it's horribly broken. anyone who can't see that has issues.

I know quite a few of people who have issues then.....

as do I. the so-called "backward reconciliation," which is tantamount to client-side hit detection, whether it technically is or isn't, makes the game nigh unplayable. some days it feels like butter, but most of the time it feels like playing in molasses. I've had games where I go 50-0 in FFA and hit 8000 damage to 1500 received, and next map vs the same group of players somehow I die 40 times with like 5000/5000. you get hit with random rails when your screen says you're ducked behind corners. you get warped into tier 2 players' plasma and when you spec them they're hitting about 12% accuracy, but somehow they kill you instantly. rockets and grenades that are nowhere near you somehow magically teleport into your face. when you want to drop off a ledge, it won't let you. when you don't want to, you slide and fall. I could go on all day. it's just broken.

What game is that? I would never play shit like that!

sounds like you are just blaming the netcode for your own poor performances and mistakes

yeah, or not.

the magical netcode boogeyman! he knows when you do and don't want to fall of ledges! by magic!

In your model would you need to lead your rail shots?

Depends...
I'd rather have the pc predict where your opponent will be. Which would mean that you won't have to lead your rail shots but he might warp depending on your ping (aka if you're on high ping he'll warp more).
But you can also choose to make him 'not warp' but then you'd have to lead your shots.


But that's just what I personally prefer... there are many other sane alternatives. It's just that the current netcode is insane for QL.

Lmao

btw there is bug with com_maxfps and sent packest/s (cl_maxpackets).
My results by perfmon.msc
com_maxfps,sent packets/s
30,30
60,62
90,91
100,100
125,125
143,84
144,84
200,100
240,125
250,125

Did you talk with sponge about this? Very strange bug, should be fixed ASAP...

Yes. He has no time for this bug. And seems like there will not be any netcode changes while sponge is in "dev team".
http://pastebin.com/KStVhq4N

You mean the dev team that creates more and more modes which aren't played while not even attempting to fix the overall gameplay to a decent level?

This reminds me of my discussions with them.... Just sad... ohhh so sad :(

you haven't seen hi-rez

You mean I can see your lg miss by miles yet deal damage on an even higher resolution soon???
omfg that would be so awesome!

was referring to the devs of tribes ascend who abandoned the game after a year

I did not know that :p

But it was already a hopeless case before release (kinda like shitmania). So I didn't follow it.

more people play tribes than quake live lol

does projectiles still account for playermovent like Tribes 2? Is skiing still taking skill to gain speed, or has it just become anyones mode of transportation, completely removing clever vehicle boosts?

In other words, is it a dumbed down tribes 2, for good or bad. Or just a dumbed down noobified version for the casuals? my intuition always told me the latter, which is why i never even tried the new one, couple with old pc ofc.

I guess what you mean with projectiles accounting for player movement is inheritance, and yes, it's still in and at 50% for most explosive weapons.

You can still vehicle boost and shrike-grabbing is still common but was nerfed a bit.

Skiing/Jetting is a bit different compared to older tribes. "Diskjumps" are a lot weaker and you have a bit of air control without using jet-pack energy for it. Coupled with health regen instead of consumable med-kits, capping/going fast becomes more a matter of time --> health --> speed: The longer you ski around, doing more diskjumps and getting multiple regens, the faster you will go. In general tribes ascend is hard to gain speed, easy to maintain. While older tribes were easy to gain speed, hard to maintain.

Just check some route videos to understand it a little better: http://youtu.be/yEL3d1gN4iQ

I think tribes ascend is a dumbed down version but it's not that bad and still a good game imo.

looks maybe a bit too fluid i guess, more like version2 of tribes 2(instead of just classic which fixed broken skiing) anyway, a middleground of ultra hardcore and a bit casual might not be a bad thing. I might have to see it pc can handle it, since it is f2p still?

Yes, it's still f2p.

i just tried it a bit, it has no heavy loadout or basecontrol with generators or anything? i did i just not see it on the test maps, instead of joining an online multiplayer on?

There are generators in CTF and heavy classes. They somewhat changed the class system so there aren't only light/heavy/medium but every armor tier now has sub classes. Heavy for example are Brute, Juggernaut and Doombringer. You need to unlock them first :/

very interesting results, it looks like netcode's work depends on local FPS :)

Not really a "bug", IMO. Maxpackets and framerate have always been linked in this way, this is exactly how it should be.

143&144fps are not valid framerate caps in the original code so it falls back to 166/2.
Same with 240fps, it just falls back to 250fps.
Same with 60, not a valid fps value so it falls back to 62.5.
If you used 142 you'd get 71 packets per second.

etc. etc.

By valid fps cap I mean a multiple of 1000, as per the original code.

They let us use any framerate value now, even ones that were not valid in the original game, but this just shows that 125 and 250fps remain the best values by far.
I bet that if you have 125/125 and your fps drops to, say, 120fps, 99fps, the packets per second will be 120 & 99. At least the last time I checked it was exactly like this.

totally right. this rtrd petka cared bout com_maxfps cap and didnt know how much maxpackets he will actually send.
even calls it a "bug" :D. rofl. stupid

Ye this is the result of the funny way in which quake is build (which also leads to this ultra responsive game when compared to other games).

This actually is not a trivial problem to solve though. (as opposed to other much more pressing issues)
Although you can just say send a packet for each frame (or if framerate is really high send 1 packet per 2 frames) and remove the command.

some players are using 126-249 com_maxfps values.
if it's impossible to send more packets at this values then they have to prevent setting this this values imho. that's why i called this bug.

it should be so nice if they would just make another thread for the networking stuff....
Would make a lot more sense in a game where fps are somewhat irrelevant. (aka it doesn't make a lot of sense if you're on low fps but on games with more than 100fps it definitely makes sense to thread the networking)

I hear you, it's a bit silly to give the ability to chose fps values that are not multiples of 1000 when the maxpackets command only works properly with those values.

But I'm just not surprised it is the case :P

wrong wording, integer values = 1, 2, 3, 4, 5 etc. (up to infinity).
So 150 definitely is an integer.

I meant integer values for 1000/fps cap. I should have slept last night , that post was a mess :p

welcome to q3 engine

Oh well, since it's out in the open now, I might as well post my graph­.

So what does this mean? If I use 250fps are there any downsides?

No. Well, none apart from the footsteps bug.
The only FPS values where you can send 125 packets per second are 125 and 250. It might be better to stay on com_maxfps 125 if you cannot reach 250 fps and want to keep sending 125 packets per second.

would be cool if someone who can effortlessly cap stable 250fps did the same measurements at com_maxfps 250, so we could see if there were in fact 125packets/s sent.

any takers? :)

Some did already. It works.

it's in cl_input.c... I don't get why people don't just check the code.
the function:
qboolean CL_ReadyToSendPacket( void ) {
......
......
delta = cls.realtime - cl.outPackets[ oldPacketNum ].p_realtime;
if ( delta < 1000 / cl_maxpackets->integer ) {
// the accumulated commands will go out in the next packet
return qfalse;
}
return qtrue;
}

This code gets called upon every frame thats drawn.
If they want this behavior fixed then just alter the above in a sane way.... I'd be in favor of removing the statement all together or perhaps if you wanna keep similar stuff going replace the 1000 with 500.

programming isn't hard :p

(edit forgot closing bracket;p)

i didn't know this part was implemented using completely identical code found in the available source, sorry :)

well it's a decent assumption. Almost all of the origional q3 code is still in ql.

Some of it is slightly altered to fix some bugs / to allow different things / etc. But in general if you think "this is bugged" you can just check the q3 code and see that it's written exactly as you thought it was.

I looked at the code quite a bit when I was in discussion with them on the netcode......

for the record.. i think they should remove the statement completely and send each frame, i doubt the servers couldn't handle the 'immense' jump in bandwidth load. christ, we have people streaming 1080p videos on youtube just to hear the music nowadays.

edit: after some digging, i think there's no point in adjusting that.. it's either leave it as it is or send every frame and cap com_maxfps a bit lower to reduce the ooh-so-scary bandwidth load.

You'll rarely see better combined code than the q3 code...
It's elegance in simplicity is beatifull hehe.

When you read it you kinda go: huh what that can't be right.... how does this even create a game?

And then suddenly it's like: omfg that's smart ;)

It's pretty much that simple :)

As nice as the q3 engine is, its a relic these days. Many of the assumptions that were made during its inception are no longer applicable. Unfortunately its all we have. I wish that id were allowed to grant a team of volunteer devs access to the source under nda to help make an amazing game just that much more amazing. xerp'd

although you're right about the assumptions.... I wouldn't see any need to update those parts....

I mean what use will it be if your eyes can't actually see the difference.

For instance these packets being sent.... that's like 8~16ms max difference if you rewrite it... That's not enough to really notice.

Similar with other things... if you rewrite the code you gain like 4 ~ 8 ms responsiveness on the screen which is usefull but not worth the effort.

(this post is not on netcode but on the general rewriting idea... the netcode seriously must change... but people know that from me ;p)

dec is right, it would be cool if they let volunteers revisit it to suit modern network and pc architecture environments better, no matter how lean and elegant it is. and you know how they say.. every 8ms is good, as long as it's free :D

well I'm all in favor of a new netcode.... but the problem is just that they added in some horrible crap.

If you use the same structure as provided you can write a much better functioning system.

But you don't have to rewrite things to suit modern network / pc architectures.

If you'd rewrite the whole of QL then maybe it'll run 25% faster with parts (perhaps) running 50% faster..... And you'd have spent more than a year on that. While atm it runs within 8ms.... So imo theres no significant value in that.

Is it dependent on the fps or the setting? If dependent on fps I guess that helps explain that lag trick. Either by screenshot or other means of dropping your fps could be exploited to send very little packets as possible(as this by passes the 30 fps limit). Could also physically remove your ethernet and plug it back again that apparently also works for temporary dropping packets.

It depends on the FPS. But then again your FPS depends on your com_maxfps unless it's set so high the limit is never reached.

As the graph shows, you can indeed send as little as 30 packets per second if you set your com_maxfps to 30. You could send even less as well if you managed to drop your FPS bellow 30 by other means. That said, I wouldn't recommend you to play on 30 fps...

lol, I prefer 100-125 in games. I'm just troubled with the fluctuating packets aspect. So in theory if everyone had X amount and went into a poorly or perfectly optimized room and everyone dropped/increased in fps, that would potentially give a spike in load. Also the exploiting below 30 fps is also very troubling. At the very least it should have a minimum value(30fps at least in regards to packets), curious how many people would start timing out frequently if that was set.

I'm not sure what you mean... what do you suggest exactly? That players from whom the server received less than 30 packets (accepted ones, so no duplicates or packets invalidated by jitter) in the last second be kicked?

*In regards to the below 30 fps problem*

Well it might cause problems, cause in its current state with different setups there might be weird abnormalities where dropping below 30 fps just happens similar to pressing screenshot. So it depends how it was implemented if poorly yeah I could see people (maybe particular people, with lossy internet) dropping quite often. This essentially has split into 2 issue's one being codependent
- packet loss
- fps dropped(artificial packet loss)

For fps dropped, essentially no one should be playing with 30 fps in the first place(my p4 computer can run at 100-125 fps). So in the event of fps dropping it should either by default send/have 30 fps worth somehow or just create dummy frames which would have a similar result.

The idea is to prevent or hinder the ability to abuse how it currently works, not kick people for lossy internet or sudden frame drops.

edit: reorganized wording/reworded

So in the event of fps dropping it should either by default send/have 30 fps worth somehow or just create dummy frames which would have a similar result.

Sending more packets than there are frames would be both overly complex and useless. I wrote a post on this very subject a few days ago :

http://www.quakelive.com/forum/showthread.php...post287228

The best solution to counter situations where too low amounts of packets reach (or are accepted by) the server would be to have the server step in and take decisions. It might make it suck for the players with such problems when the decisions are far from the reality but whatever, it should be their problems ONLY.

Complex yes, I don't know how the engine works so I'm just throwing out ideas. I agree with the server stepping in for this particular case.

But as I mentioned this seems to have split into 2 situations, would it be possible to still meet packet requirements but only send the interesting bits by manipulating fps? For example have 200 fps, drop to 30, and miss one and a half of a second of possibly important information.

No, the same information is transmitted, no matter how many packets you send. Typically you're gonna send larger packets if there are less of them (it's a delta between the current state and the last acknowledged state by the server that you know off). The problem isn't the information transmitted but the gaps in time between the discrete receptions of it by the server.

your packets are so fat that when you jump, you've already landed on your client side before sending :)

ain't gonna work....

You'd end up with a type of movement for people on low fps which is much much worse to cg_noPredict 1.

It's not like the problem is in sync..... There is this famous thing called ping.

Do we actually know just how low the (accepted) packet reception rate on the server has to be to have people noticeably warping? I doubt anyone with more than 40 FPS would cause real issues unless he happens to also have connection issues. Low FPS isn't really the problem here IMO.

It's true that if the server had to make predictions and set them as "the truth" for the other clients it would have to affect the problematic client as well in a fashion similar to local prediction errors but worse because of the latency... but I think it's better to confine the problem to the bad client than to force everyone to deal with it. It might not turn out that bad anyway. I'm not sure that much happens in a typical ping/2 delay that it would make the game unplayable for ex-warpers. I'd be curious to try it.

Well there are tons of mis-understandings in this field of research. There are a lot of researchers saying "If we show some usa jet pilot an image for just 1ms they can tell what it is". Which makes people think that you can see things in 1ms (which is utter bs). Those things are always done in dark rooms with very bright images. (Which means that your eyes first adjust to no input so chemically they become neutral, and then you flash something really short at them which makes the chemicals react -> lead to an image.... But to see fps is very different as your brain actually updates parts of the image out of sync with the input etc.)

Also the term "warp" is rather vague.

In general it's really hard to test though... because our own eyes trick us quite badly.
If you expect something to "not warp" your mind will do the same thing that QL does... interpolate between states.

In general though if someone has 40fps and is updating the server on a stable connection each frame. Then you'll not see him warp at all. Because the server runs at 40fps.

But the question is how "accurate" the description of the server then is.

For there is quite a large difference for the accuracy in between recieving the snapshot 1ms before the "new state" is calculated vs 1ms after the "new state" is calculated (23ms difference). I'm not sure how the current netcode deals with this though. But the concept is quite obvious.
On higher fps this difference becomes neglible and it gets negated through a random factor (aka on 120fps max 7ms inaccuracy vs 1ms inaccuracy -> feeling 4ms off is accurate).

However on 39fps you'll have "occassional warp" because then the difference will of accuracy will jump between 48ms and 1ms. (once in every 25 seconds you'll "miss" informing the server before the next snapshot is made)
On 38fps that's 2ce in 25 secs... on 37fps it's 3x in 25 sec.


This is also why even a slight hickup in your connection can lead to just missing the next snapshot -> warp on the nmy pc. While the warper doesn't notice it at all.
In general there are 3 server updates in 1 snapshot (120 fps) so when there are just 2.... well you can guess :x

This problem remains though but how you solve it leads to different positives/negatives.

To add on that btw.... The really old films are actually "quite vague". That is, if you take a movie still (cut out 1 of the images of the film roll) all movement is blurred... and often you can't even make out what happens if you look at 1 image.
But if you put them in sequence your mind will interpret the "vague input" as a constant input for XX ms what would in the real world be seen in the way but over time. So they would appear to be sharp and extremely fluent... even on 25fps.

The only "warpy" part is when the next image (from the roll) was loaded. The camera would "shut" and load the next image then open to record and then shut again.
The time it was shut -> how warpy it is. Rather than the number of frames produced.

This is also why the "early" animated movies (which are all generated purely sharp images) gave people a headache... And why we're so obsessed with framerate nowadays.


If you think about it it's actually really logical... but it's hard to explain hehe :p

This will never ever be implemented. The levels of packet loss on your average bum's wireless connections are too high, and devs want to keep the game playable by every kind of peasant.
Quake is one of the few games where this would have much of an impact, so unfortunately we have to keep copping it.

By the way, amazing work noticing that!

why didn't you include com_maxfps 333, any obvious reason i'm missing here?

Yes : it's capped at 250 in QL.

Thats how its supposed to be.

125fps ftw

Big thanks for bringing that out. Recently I found out that my LCD can run at 144 Hz so I moved to com_maxfps 144 to have better sync. I didn't know that it affected connection responsiveness in such way.

Now I'm going to experiment with these values to find the best suite. I can't reach 250fps on my PC so I would probably go for com_maxfps 125 and r_displayRefresh 144. It should be still more smooth than 125fps/120hz in case of video rendering.

No, because :
1. com_maxfps 126 is exactly the same as com_maxfps 142. To really test 126 I would have to manage to get stable on 126 which would be tough if not impossible.

2. I can't even reach 230 FPS. com_maxfps 230 is the same as 250 anyway.

EDIT : Well here's the post I was replying to before it disappeared :

Could you please do a test for me to see how many packets are sent per second on 126 and 230 fps?

sorry i just realized that was the case after i posted it

but ur graph shows it sends 92 packets a second on 250fps and 124 on 125?

edit: oh wait, fraps fps :/ fail

id Software is looking for a Senior Networking Programmer to work on...
http://jobs.zenimax.com/requisitions/view/624

Good organization and time management skills

not for QL though

Hmmmmmm..... That would be.... interresting to say the least :D

The game's netcode is much better than people give it credit for. Try playing old osp or even cpm, its netcode is dreadful in comparison as are many other games.

even cpm?

Ye

cpm depended fully on your and your opponent settings.... further on people, in general, settled for a certain setting which played pretty good (and was not an illogical choice).
If however you did not use those settings... well you could be under the impression that it was bad ;p

Compared to lan it felt like absolute shit no matter what settings I configured >_<

every netcode feels like garbage compared to lan.
It's because there are obvious problems in transferring data which you cannot ignore.

ever played loadout/firefall?

firefall I can't tell if I'm playing on 100 ping or 20

Plays fine at 200 too

loadout I haven't played above 100 but even then it feels the same as 20

Quake live has a different feeling between 20 and 40


QL netcode is garbo

client side hit detection, essentially a free pass for cheaters.
Would be absolutely horrible in quakelive as it would allow for for people to wreck their own connection(warp all over the screen) and still hit every single shot.

If you could somehow brainwash people into not being assholes it would make sense.
FireFall when it was still active was ridicilous because of this.
Loadout will suffer the same given enough time.

quake is suffering the same right now....
More and more people are wrecking their own connection in order to win games.
Now a few years ago this only happened unintentional.... but lately people are chosing to play on wifi over plugging in their cable.

Im speaking of more extreme cases, having a shit ton of packetloss and warping/jittering all over the screen while still being able to hit 40-50% lg would essentially make the game unplayable for the opponent.

While at the moment it makes it annoying you are still able to adapt to it.

I am not able to adapt to it. :(

are you kidding? if you are going to hack you are going to hack regardless

whether or not it is clientside or serverside makes no difference, you will get banned either way
it should be conditionally client sided like loadout
not like tribes ascend where it's broken

What kind of backwards logic is that?
Your essentially giving cheat creators direct control, having the server verify things helps to remove the ability to kill everyone on the server but does not fix the issues with aimbots.

Also see the difference between a game like loadout and quake.
The majority of combat in loadout is in open areas while combat in quakelive is more about angles,positioning etc.

A better example is to play bf3 on smaller maps against people with high ping.
"whether or not it is clientside or serverside makes no difference, you will get banned either way"
is that why loadout is running rampant with cheaters?

What cheaters? People hackuse me nonstop from my first 5 hours of playing and I'm not even very good at it yet. Some "pros" in one of my other loadout videos called me having an autofire and then their whole team starts hackusing me. (This is not the videos below, that one got expired)

http://www.twitch.tv/jigglywigglyx/b/504229077?t=16s
and even 1m46s in that clip, wau det autofire
http://www.twitch.tv/jigglywigglyx/b/504229077?t=25m30s

Loadout is not 100% client side, it's a balance.

You afks are talking about the game and haven't even played it. Of course most quakers are awful at the game because it favors some mousing over plain garbage 30% lg waggle like zero4.

So your argument for people not cheating in Loadout is that you get accused alot..... Awesome

Loadout uses clientside hit detection with server side hit verification..
What method of verification they are using i dont know, ask the developers.

"You afks are talking about the game and haven't even played it. Of course most quakers are awful at the game because it favors some mousing over plain garbage 30% lg waggle like zero4."

It took 2 comments of me stating facts about clientside hit detection and the problems it poses to you saying we are garbage.
You my friend are an asshole.

ofcoz

It's called detecting for injects.
Also there is a limit to the client differentiation between the server and client.

People love to hackuse in that game so no, I don't really believe that there are any rampant hacks. gst800 has been called a hacker on literally every game he's played when watching his stream and they all report him.

Cheats are widely available, if you choose to believe people use them or not is completely up to you.

It's also hilarious you assume no one else on here has been cheat accused before.

goodnight

considering 90% of the ca players think I cheat(including extremerz) then my point stands.

I haven't seen anyone get hackused more than me in ql NA besides dynamicmode.

even fatal1ty called me a bot

hackusations mean nothing at all, and if someone is hacking on serverside/clientside makes no difference

bots are so easy to see
you simply look at their mousing when they aren't shooting someone vs when they are

Then why keep bringing it up...

because you said client side hit detection is a free pass for cheaters

cheaters will get caught regardless, and hackusations from that forum post mean nothing

1. Should have rephrased that a bit, meant the authors of the hacks not the cheaters themselves.

2. I didnt give you an example of an accusation, you dragged it up and tried using it as evidence of the contrary even though there is plenty of proof to support my claim.
Go look for yourself.

If they get caught or not doesnt matter, 15years of counter-strike should be enough proof of that.

https://loadout.com/forums/index.php?threads/...-you.3936/


Does not sound like fun except for hackers.
Except you can't fairly ban them unless you can prove they're tampering with their bandwidth intentionally.

No thanks.

I have not played those games.

at least play loadout the netcode is so good
and so is the game

gst800 who posts here plays and got a nice twitch following because of it
http://www.twitch.tv/gst800/b/499884248?t=47m43s
loadout is full of one shot sniper shots and there are so few "non-regs" in that game

the balance between your shots counting and being shot behind cover is extremely good

in quake live your rails can even shoot through the structures completely sometimes(smooth walls disabled)

Most of these games have the benefit of having much slower movement, which allows the interpolation to be more accurate and responsive at higher latency. As well as the built in lag having less of an effect. As well as many games featuring much more integrated clientside detection systems.
Quakelive, to the credit of the developers, has probably got the most difficult challenge for a network engineer. It's a lightning fast game, where accuracy is important in enclosed spaces.
My only real issue with QL's netcode is the tolerance levels for packetloss, the lack of transparency to show who is losing packets, and the way the game performs in these adverse conditions. Also, the poor condition of servers and how accurately they deal with backwards reconciliation.
Has any developer thought been given to running less idle pub servers? And implementing a system allowing free users to initiate these servers? instead of instaFT servers running idle and empty all day, being polled by the server hardware when they contain 0 players for 3 weeks at a time...

Quake live is not that much faster
what difference would have clientside hit detection for lg do? It would let the better player win obviously. LG has no reason to not be client sided.
the benefits of an acceptable clientside hit detection(conditional, not pure client side) far outweight the negatives of a poorly implemented server sided one
with plenty of interpolation to stop people from stuttering the netcode would be way better.
e.g players like isbadmuba who just warp would be smoothed out
in ql people who have 5ms ping jitter warp a lot and 5ms of jitter in new release games is completely interpolated and smoothed out generally

of course this would require a complete redesign
i do not think the ql devs have the talent for this, needs some mmorpg devs

just watched a minute of a gameplay clip.... Loadout is not even 10% of the speed of quake. The models are much larger the areas you fight in are much larger the explosions are huge.
Now to you that might not seem as though they alter the speed of the game. But they do.
Also there aren't many straight edges on the map which I saw which, combined with the large models, leads to you thinking that you never get hit behind a wall.

Also there are many reasons why clientside hit detection is bad. And none of them have to do with warping being smoothed out?
Also the only way to reduce warp while interpolating is to let the server send less snapshots to the clients. (effectively giving everyone a worse experience)

What you can do is 'interpolate between extrapolated states' but that would still be called extrapolation. (Aka the "real model" warps but you draw it as if it's fluent from one warped state to the next)

10% of the speed of quake live... how?

you realize loadout is full of one hit kills with the snipers? gst800 was sniper 1v1 dueling boss(one of the best sniper duelers in loadout) and their ping differentiation was huge. They were cover spamming and they didn't die behind cover much at all. There was a 160 ping difference and the game felt fine. This is a headshot duel behind cover spamming.
A blend of client/server netcode favors aimers far more than complete server sided nonsense ql has. i.e the noaim scrublords who said BC2 netcode is better than BF3.

You also realize the netcode could be tweaked from loadout so that it would be less leniant.
I understand the shooting mechanics of QL well enough to realize that the netcode is completely outdated.
Gettin killed slightly behind cover happens anyway with QL's atrocious netcode.
Also you don't understand how QL works. How much cover rail cover shooting is there in the game? Rail isn't a one hit kill weapon and almost all the duelers avoid these situations except a few like tox/spart1e. Most duelers just spam rail without looking, how would that affect this at all?

But it's pretty pointless to argue with someone who's only fps experience is quake live or games of that era. They have absolutely no idea what they are talking about. They have no experience with any new fps and how they handle the netcode.

They have also never played games outside the genre e.g mmorpgs/fighting games.

If anything to make both sides happy, rail should be mostly server sided and lg mostly client sided. Of course that makes way too much sense.

Oh, but wait I'm supposed to lose an lg duel if I have 110 ping and some mediocre lg'er with 10 ping should beat me... right

They have 10ms jitter? They better warp all over the place...

I am unable to give a sane reply to such an insane post.

Please get a clue on what latency is and understand that there cannot be any 'great solution'.
Every(!) sane solution has an upside and a downside compared to other sane solutions.

This is inherently true because you cannot instantaneously transfer information.


This means that you cannot make the game feel the same for ping 110 as it is for ping 10 unless it is insanely slow.
That loadout manages to do this is because the models are huge and the levels even larger.
If you would have close combat fights and watch both screens you will see shots which blatantly miss on one screen hit on the other.
(unless the movement is extremely different from quake (it would feel like 'cg_noPredict 1' in QL although perhaps smoothed out))

Personally I can't stand QL's backward reconciliation when playing ranges of pings. It's only ok when everyone's on about the same.

clone a player and have one on 10ms and one on 40 or 100 and yes the 10ms player should always blitz them, the same as someone on a £150 p3 at 60fps should be at a disadvantage to someone on an i7.

Sure, do what you can to make it as playable as possible for the disadvantaged, but it should never be at the detriment of superior technology. QL's code just seems to have gone to far somehow.

You shouldn't expect preferential treatment if you try run a fiat panda in an F1 race.
Everyone would be pretty pissed off it they were all limited to 50mph.

For the bulk of my Q3 time I had 110ms at best, I just accepted that to beat people on much better connections I had to drastically change tactics, hide a lot more, predict tons. Now it just seems barreling in is the way to go because chances are you're going to be 3x harder to hit than a regular player, but you have only maybe half the chance to hit the same accuracy as you would on a better connection.

funny how you call it insane

for an lg duel if I am hitting him on my screen my lg should count if it's within reason
it should not simply be unplayable
i don't know why you are talking about netcode when you have never played a new fps.
explain what the negatives are for a favored client side lg

note I am not saying 110 ping should feel the same as 10, I am saying it should be a lot closer in feeling. feeling the same would be pure client side which is abusable if your net is really bad(full of jitter)
overly sided client compensation become a problem
also overly client sided hitdetection are a problem for millitary shooters
if you get quick-scoped and you die behind cover that is far more frustrating than anything in QL
QL is far more open combat than military shooters

fast paced has nothing to do with the issue at all
the only reason it can become a problem is for cover railing adadad spam but the rail reg is broken anyway. so making it favor aimers is far more beneficial than the current state

If you see your opponent aiming at you, and you dodge to a side and see the shot pass you on your screen.
Should you get hit?
And at what area is "within reason" for that?

I've studied the subject of synchronising systems for about half a year or so on the university (it was contained within some of my regular courses). I don't need to know how a specific game solved it. I simply know all possible solutions to the problem.


The negatives to a clientside lg is that your own position is extrapolated (further than what the server knows) on your screen while your opponents position is interpolated (behind what the server knows). This (indirectly but inevitably) means that it is possible to hit lg on your opponent without him ever seeing you. This effect gets enhanced by having a bad connection yourself. And thus people will be more and more tempted to play offensive with lg while on high ping. (kinda like QL has become.)

edit (add small example): If you move into a room on your computer, you will imediatly enter that room on your computer. This will take a while before that happens on the server, and it will take longer before it can happen on your opponent's screen.
Vice versa When your opponent leaves a room you will be informed of this much later than when he left it on his screen.
Therefor if you would (on the server) enter and leave the room simultaneously. On your screen you'll have a constant factor * your ping time to aim at the opponent. While your opponent will have MINUS "that same time" to react. (so his reactions must be minus a factor of your ping. And to react on something before you know that it took place... well that's impossible would you not agree?

i don't care what they are hitting
i can outdps anyone except like extremerz if he's in shape
i don't care where they are aiming and neither should anyone

Here is quake live netcode at the moment:
1. Huge difference between 10 ping and 60
2. The player with lower ping will will if they are a worse player.
3. No international gameplay.
4. Rails constantly go through people because the hitbox is delayed.
5. When people warp they start jittering all over
6. Even if you are shooting at the warped model it still often does not reg.

Here is much better netcode:
1. Client side favored with checking (60 tickrate+) that counts everything you hit on their playermodel. It will not count if your values differ too much from the server based on your current location(movement where you are relatively)
2. Through rails are eliminated
3. Interpolation/Extrapolation of player movement so they do not warp all around.

This netcode favors aimers FAR more and is a significantly better tradeoff.

About your point on making people play more aggressive with LG, that's a good thing. You are already seeing them come delayed at you with the server sided netcode. The time it takes for someone to react is an average of 215ms+ (250ms) on the new statistics. That is slow as hell
http://www.humanbenchmark.com/tests/reactiontime/

I can do 135ms, with that I am approximately 80ms faster than average. 80ms, that's a huge amount of time.

If people react on average of 215-250ms, then surely a ping difference of 40ms in QL's netcode shouldn't be so drastic.

Right.... so the only thing that matters in FPS is how you aim. To win you go head to head and the most damage per second wins!
That it becomes impossible to dodge things doesn't matter. Only hit detection matters babeh!

Anyways.... your global idea of the problems due to latency are completely fubar. Please study the subject before thinking you're an expert because your reaction speed is better than most.
In formula 1 the driver doesn't design the car because he's a better driver than the designers.

Even the other people in this thread have pointed out that loadout has seriously flawed netcode.



btw: "Interpolation/Extrapolation of player movement so they do not warp all around." <-- you have no clue what that means....

And now you all have the opportunity to fix it yourself!
How cool is that?

How to win at QL?

Maxfps 144 + Wifi + torrents

the netcode is perfect compared to the sound.
you need to play maps 100 times to be able to guess where the enemy is from sound alone.
i just played cure, which i did maybe five times before, and while i was constantly hearing the enemy, i still had no clue where he was.

it's such a shame that this otherwise very accurate game, has this crap soundengine.

They dumbed a lot of the game down.


......................
Can hear more (dm6... can hear Rg launch pad from rl lol)
Hear further ( hear whats on the other side
of the map when you shouldn't be able to)
Power ups glow through walls lol

i think you can position yourself by the RL and hear the RA teleporter on furious heights :D

You can. When I started playing that map I used to think an opponent sneaked past me and took the mega-tele. Was confusing!

I think audio positioning is fine, it's when sounds don't play that it gets irritating...

I rage when a player use their warping to their advantage. I'm not suggesting they do it on purpose, but it's just a skill they have developed over the years since it works. I know a few people who run around +forwarding like man men, rocketing and sometimes lging people. But you can't hit shit because they flicker like crazy. They think it's all in their 'dodge'.