Im just wondering this too... To be sure that Im not halucinating; I checked both versions net settings and played couple of beta4a and beta5 rounds in a row... There is a significant difference between them... which eliminates the steam side...
There is a problem; and probably because of netcode tweaks done in beta5...
Ummm... also there is no word from any dev yet... Probably they'r checking the code to be sure... <!--emo&:D--><img src='http://www.unknownworlds.com/forums/html//emoticons/biggrin-fix.gif' border='0' style='vertical-align:middle' alt='biggrin-fix.gif' /><!--endemo-->
I dont even bother to read this whole thread, but yeh, its worse... Atleast I've had several encounters where skulks seem to sustain more bullets, even with goo all over the place.
I didn't read the whole thread but i also had the same problems. Especialy with leaping skulk and celerity lerk. Skulk now sometimes takes 2/3 sg hits to kill and lerk die when i shoot behind them.
<!--QuoteBegin-Emanon+Sep 10 2004, 01:23 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> (Emanon @ Sep 10 2004, 01:23 PM)</td></tr><tr><td id='QUOTE'><!--QuoteEBegin-->Yumosis, I understand why this is happening and I'm willing to accept it(it is a old engine). What I cant understand is why there is such a difference between beta 4a and 5. From what i can recall there were no speed changes in the classes. Something someone tweeked somewhere is hurting gameplay.<!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd--> It's possible you're just noticing it (the better a player becomes the more problems he will knowtice with the game), or it's possible server you may be playing on is getting poor FPS(less than 90-100). A server will view things like this <!--c1--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1--> Server with poor fps X X X X X X X X <!--c2--></td></tr></table><div class='postcolor'><!--ec2--> <!--c1--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1--> Clients with good FPS YYYYY YYYY YYYY YYYYYYY YYYYYYY <!--c2--></td></tr></table><div class='postcolor'><!--ec2--> With cl_cmdrate you update based on the server or client fps, so, we have 100 fps and a cmdrate of say 100, but the server is only getting 50 fps, we get what's called choke, we are sending more in than the server can see, or the otherway around if we are getting poor fps and using a cmdrate higher than our fps.
As far as anything being changed netcode wise in beta5, I don't think it was even touched(maybe a dev can answer this), <i>but</i> it's possible the server you often play on is getting worse fps than it was in 4a, for maybe something as simple as the new dropped ammo.
However, it could even be a problem with the way ex_interp is being calculated, similar to how ex_interp use to work in WON, but I don't think it was changed, just unlocked. (again a good question for the devs)
If you have rcon to a server, you should try sv_maxrate 25000 and sv_maxupdaterate 100 and use cl_updaterate 100 and rate 25000 to see if it makes this issue better. If it does, you are just knowticing a common problem that's been around forever in ns.
I understand your conprehension on the netcode but let us discuss something else.
From what i understand the server will only register hits it knows you were shooting. Everytime your computer sends out a signal to the server saying "i am shooting" the server checks the location and confirms it. What my point is that even though moving targets are faster in ns, client side prediction is not as horrible in my opinion as one would make it out to be. With an update of 20 correctly aligned frames with the server one *should* be able to hit the skulk.
Now here i get into something i do not understand completely and want to discuss to prove or disprove.
Even though the server can only update you for example 20 times, does that mean you can not update the server 100 times? For example i noted with a ex_interp .05/cl_updaterate 20, some of my shots would disappear. I would shoot the skulk and yet the skulk would keep moving at me. The bullets would just be lost.
Yet with ex_interp .01/cl_updaterate 100 if i shoot a moving skulk with the proper amount of bullets(accurately) it will die.
So let us assume this: If you set your client to update the server 100 times a second(Only possible with 100fps) then every bullet you fire it will send a signal to the server i am shooting. The server then takes the 5 signals you sent out in that time and responds with 1 signal confirming or not the hits.
Now the other issue people also have is they say cl_updaterate 100 is for lan. If your client updates the server 100 times a second, what i understand of it is it will send out the signals a 100 times in a second. That each of those signals will take 40-80 ms to reach the server is beyond the point.
All the testing I have done is on a servers that I run. Constant fps of 150 and no network restrictions. I ran the server during 4a and 5. So it isnt the server. (same server.cfg btw) So no variables have changed besides the beta versions
sv_maxupdaterate 101 sv_minupdaterate 30 sv_maxrate 30000 sv_minrate 5000 sys_ticrate 150 FPS is between 120-150
Very interesting, this very could be a real bug of some kind, but looking through the change log I see nothing that could be causing a bug like this, the only thing remotely close that could be causing this is ex_interp(similar to the way it use to work on WON, made the hitbox lag behind). Maybe something was left out of the change log, guess we'll just have to wait for a dev to comment. :/
I found that if I leave all my settings at default (I dont touch ex interp, I dont touch cmdrate, I dont touch uprate, or anything... the only things I touch is rate and fps max (100)), I hit pretty much exactly where I'm pointing every time...
Perhaps messing with the console vars was a bad idea.
No problems with the new netcode for me, although last inventory deosn't work well, i could care less as i use the 1,2,3,4,5,6 keys
edit: (did they change the netcode atall, i can still shoot skulks fine)
(4 being welder, 5 mines, and 6 handgrenade, bound seperately)
althought a cancel reload script i made using last inventory didn't work anymore, I heard this is the cause of sometimes not spawning with certain weapons, or last inventory or slots not working is the method of fixing certain blink/leap exploits
I have been noticing this problem recently as well and now during games, i have now turned on cl_showerrors to see what the prediction errors... I must admit I never played b4a with it running while playing a normal game but reviewing it over in the height of combat NS or CO the prediction errors go through the roof... Also another thing to note.. When walking normally or checking corners occasionally I experiance a slight spike in prediction errors as well as having a constant prediction error of 1. (5 instances of it.)
Might I suggest people here do the same and monitor it?
Agreed, I can't say that I notice it with LMGs much but it is fairly noticable with the shotgun.
At least it's not anywhere near as bad as CS, I don't have any clue what's going on there.( Just switching servers(people seem roughly as skilled) makes a night and day difference, going from people calling me "OMG nub" to people calling me "OMG hax" and "WTH! BS!" <!--emo&???--><img src='http://www.unknownworlds.com/forums/html//emoticons/confused-fix.gif' border='0' style='vertical-align:middle' alt='confused-fix.gif' /><!--endemo-->).
edit: We need something like a command for displaying roughly how accurate hitprediction is(which we can do, since the client does predict hits, we can count these. The server also calculates hits, we only need a method to check for how many hits happened server side on the client. If the server directly gives you damage and inflictor this isn't that hard to do.). The shotgun has a static spread on the server but random on the client, this would allways generate some errors where the client thinks it hits but the server thinks it misses. LMGs are random on both client and server but I don't think the random seed is shared( if it was it would make hit decals more accurate, but since so many bullets are needed it's safe to say that on average it's going to matter very little). The price for using a shared random seed is seen clearly in CS, where anti-recoil cheats exists. So the best you can hope to do is assume that on average there is supposed to be as many hits and misses on the server and client and use the procentual difference as a meassure of how bugged hit prediction is.
It would be even better if this was done with a client-hook(even if this is the way cheats do their dirty bussiness, it's how you put it to use that matters, as long as it can't be abused. Closed source of course, even better if it's just made available to PTs and stays that way). Then you could directly compare any version of NS with any other.
I have noticed a difference, especially with the shotgun. Many skulks end up requiring multiple blasts in this beta while previously they may only have required one under the same circumstances (same upgrades, etc). Coincidentally, it doesn't seem to be a problem when the skulk is not moving.
Also, as a skulk, it feels like I should be dead before I reach the marines, yet I can usually approach and get a few bites or a kill in. It's strange when you're a marine and skulks say, "Wow, I should have been dead," instead of "OMG YOU KILLED ME WTH HAX!".
Some dev input on this would be greatly appreciated <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html//emoticons/smile-fix.gif' border='0' style='vertical-align:middle' alt='smile-fix.gif' /><!--endemo-->
<!--QuoteBegin-milosis+Sep 10 2004, 06:37 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> (milosis @ Sep 10 2004, 06:37 PM)</td></tr><tr><td id='QUOTE'><!--QuoteEBegin--> every server i play on, i have to aim behind the skulk to hit anything. against every single skulk.
oh and btw Yumosis is god. <!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd--> Seems I can confirm this. I lowered my sensitivity just enough to lag behind the target a bit - and it gets alot easier all of a sudden :o (But once they start going up walls shooting straight at them and slightly behind them both seems to produce equally ineffective results - to the point where I can safely say its not ALL just me missing <!--emo&;)--><img src='http://www.unknownworlds.com/forums/html//emoticons/wink-fix.gif' border='0' style='vertical-align:middle' alt='wink-fix.gif' /><!--endemo--> ).
<!--QuoteBegin-Emanon+Sep 11 2004, 10:21 AM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> (Emanon @ Sep 11 2004, 10:21 AM)</td></tr><tr><td id='QUOTE'><!--QuoteEBegin--> "interp" not ex_interp was by default off.
When I turn it on it says: "interp 1"
Frame Interpolation ON
Anyone care to help investigate wth this thingy does? <!--QuoteEnd--> </td></tr></table><div class='postcolor'> <!--QuoteEEnd--> Interp only smooths the animation of sprites, nothing to do with model interpolation :/
I've been noticing this bug since beta 3.0 began. Are there any ways to fix the rates so that it's easier to hit things? I'd tried before to lower my ex_interp to .01 or whatever but all that did was make the models horribly jerky. I use a rate of 20000 with cl_updaterate and cmdrate at 101 I believe. Any suggestions?
ssjyodaJoin Date: 2002-03-05Member: 274Members, Squad Five Blue
edited September 2004
there is a major problem.. im doin a hell of a lot worse than i ever did... those damn skulks get hit over and over and dont die... i blame it all on ssjyoda. lets start an anti ssjyoda petition and get him off ns. heh $1,000 to the man that brings me his head.
Well that's a witch hunt if I've ever seen one, ssjyoda.
Two things that have definately changed since b4a are ex_interp and the movement code. ex_interp being set incorrectly compared to your other rate vars could easily cause misplacement of models on the client, but the default hasn't changed, so if you haven't messed with it then this isn't the problem. That doesn't mean that fiddling with it can't improve the prediction.
I'm more suspicious of the changes in movement code. Sometime before b5 was released, I read on Mantis that the fixes for jetpacks were causing problems with the other forms of movement. While the main problem has certainly been fixed (and thus we have b5 to play with), that doesn't mean there isn't some residual problem.
I havent noticed any problem with hitboxes for a long time. I'm pretty good at shottying skulks now (and biting jetpackers, which is same thing really, since they move just as fast).
Note though that I keep all my settings at default.
It may very well be that messing with your settings like ex_interp and updaterate and cmdrate are screwing you up. It might be best to leave them at defaults. That way you see the same thing that is on everyones screen, and it is how it was meant to be.
<!--QuoteBegin-Emanon+Sep 9 2004, 07:37 PM--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> (Emanon @ Sep 9 2004, 07:37 PM)</td></tr><tr><td id='QUOTE'><!--QuoteEBegin--> What I have tried is various combinations of the following: Default ex_interp and connection settings. Then ex_interp all the way from 0.1 to 0.03 cl_cmdrate from 50-100 cl_updaterate from 50-100
Thanks, Emanon <!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd--> I would just like to remind you.
Forlorn, did you adjust any settings or just get used to where you need too shoot?
Comments
There is a problem; and probably because of netcode tweaks done in beta5...
Ummm... also there is no word from any dev yet... Probably they'r checking the code to be sure... <!--emo&:D--><img src='http://www.unknownworlds.com/forums/html//emoticons/biggrin-fix.gif' border='0' style='vertical-align:middle' alt='biggrin-fix.gif' /><!--endemo-->
Atleast I've had several encounters where skulks seem to sustain more bullets, even with goo all over the place.
It's possible you're just noticing it (the better a player becomes the more problems he will knowtice with the game), or it's possible server you may be playing on is getting poor FPS(less than 90-100). A server will view things like this
<!--c1--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1-->
Server with poor fps
X X
X X
X X X X
<!--c2--></td></tr></table><div class='postcolor'><!--ec2-->
<!--c1--></div><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>CODE</b> </td></tr><tr><td id='CODE'><!--ec1-->
Clients with good FPS
YYYYY
YYYY YYYY
YYYYYYY YYYYYYY
<!--c2--></td></tr></table><div class='postcolor'><!--ec2-->
With cl_cmdrate you update based on the server or client fps, so, we have 100 fps and a cmdrate of say 100, but the server is only getting 50 fps, we get what's called choke, we are sending more in than the server can see, or the otherway around if we are getting poor fps and using a cmdrate higher than our fps.
As far as anything being changed netcode wise in beta5, I don't think it was even touched(maybe a dev can answer this), <i>but</i> it's possible the server you often play on is getting worse fps than it was in 4a, for maybe something as simple as the new dropped ammo.
However, it could even be a problem with the way ex_interp is being calculated, similar to how ex_interp use to work in WON, but I don't think it was changed, just unlocked. (again a good question for the devs)
If you have rcon to a server, you should try sv_maxrate 25000 and sv_maxupdaterate 100 and use cl_updaterate 100 and rate 25000 to see if it makes this issue better. If it does, you are just knowticing a common problem that's been around forever in ns.
I understand your conprehension on the netcode but let us discuss something else.
From what i understand the server will only register hits it knows you were shooting. Everytime your computer sends out a signal to the server saying "i am shooting" the server checks the location and confirms it. What my point is that even though moving targets are faster in ns, client side prediction is not as horrible in my opinion as one would make it out to be. With an update of 20 correctly aligned frames with the server one *should* be able to hit the skulk.
Now here i get into something i do not understand completely and want to discuss to prove or disprove.
Even though the server can only update you for example 20 times, does that mean you can not update the server 100 times? For example i noted with a ex_interp .05/cl_updaterate 20, some of my shots would disappear. I would shoot the skulk and yet the skulk would keep moving at me. The bullets would just be lost.
Yet with ex_interp .01/cl_updaterate 100 if i shoot a moving skulk with the proper amount of bullets(accurately) it will die.
So let us assume this: If you set your client to update the server 100 times a second(Only possible with 100fps) then every bullet you fire it will send a signal to the server i am shooting. The server then takes the 5 signals you sent out in that time and responds with 1 signal confirming or not the hits.
Now the other issue people also have is they say cl_updaterate 100 is for lan. If your client updates the server 100 times a second, what i understand of it is it will send out the signals a 100 times in a second. That each of those signals will take 40-80 ms to reach the server is beyond the point.
Therefore do you agree or disagree and explain.
sv_maxupdaterate 101
sv_minupdaterate 30
sv_maxrate 30000
sv_minrate 5000
sys_ticrate 150
FPS is between 120-150
Perhaps messing with the console vars was a bad idea.
edit: (did they change the netcode atall, i can still shoot skulks fine)
(4 being welder, 5 mines, and 6 handgrenade, bound seperately)
althought a cancel reload script i made using last inventory didn't work anymore, I heard this is the cause of sometimes not spawning with certain weapons, or last inventory or slots not working is the method of fixing certain blink/leap exploits
Might I suggest people here do the same and monitor it?
At least it's not anywhere near as bad as CS, I don't have any clue what's going on there.( Just switching servers(people seem roughly as skilled) makes a night and day difference, going from people calling me "OMG nub" to people calling me "OMG hax" and "WTH! BS!" <!--emo&???--><img src='http://www.unknownworlds.com/forums/html//emoticons/confused-fix.gif' border='0' style='vertical-align:middle' alt='confused-fix.gif' /><!--endemo-->).
edit: We need something like a command for displaying roughly how accurate hitprediction is(which we can do, since the client does predict hits, we can count these. The server also calculates hits, we only need a method to check for how many hits happened server side on the client. If the server directly gives you damage and inflictor this isn't that hard to do.). The shotgun has a static spread on the server but random on the client, this would allways generate some errors where the client thinks it hits but the server thinks it misses. LMGs are random on both client and server but I don't think the random seed is shared( if it was it would make hit decals more accurate, but since so many bullets are needed it's safe to say that on average it's going to matter very little). The price for using a shared random seed is seen clearly in CS, where anti-recoil cheats exists. So the best you can hope to do is assume that on average there is supposed to be as many hits and misses on the server and client and use the procentual difference as a meassure of how bugged hit prediction is.
It would be even better if this was done with a client-hook(even if this is the way cheats do their dirty bussiness, it's how you put it to use that matters, as long as it can't be abused. Closed source of course, even better if it's just made available to PTs and stays that way). Then you could directly compare any version of NS with any other.
Also, as a skulk, it feels like I should be dead before I reach the marines, yet I can usually approach and get a few bites or a kill in. It's strange when you're a marine and skulks say, "Wow, I should have been dead," instead of "OMG YOU KILLED ME WTH HAX!".
Some dev input on this would be greatly appreciated <!--emo&:)--><img src='http://www.unknownworlds.com/forums/html//emoticons/smile-fix.gif' border='0' style='vertical-align:middle' alt='smile-fix.gif' /><!--endemo-->
oh and btw Yumosis is god. <!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd-->
Seems I can confirm this. I lowered my sensitivity just enough to lag behind the target a bit - and it gets alot easier all of a sudden :o (But once they start going up walls shooting straight at them and slightly behind them both seems to produce equally ineffective results - to the point where I can safely say its not ALL just me missing <!--emo&;)--><img src='http://www.unknownworlds.com/forums/html//emoticons/wink-fix.gif' border='0' style='vertical-align:middle' alt='wink-fix.gif' /><!--endemo--> ).
When I turn it on it says: "interp 1"
Frame Interpolation ON
Anyone care to help investigate wth this thingy does?
When I turn it on it says: "interp 1"
Frame Interpolation ON
Anyone care to help investigate wth this thingy does? <!--QuoteEnd--> </td></tr></table><div class='postcolor'> <!--QuoteEEnd-->
Interp only smooths the animation of sprites, nothing to do with model interpolation :/
Two things that have definately changed since b4a are ex_interp and the movement code. ex_interp being set incorrectly compared to your other rate vars could easily cause misplacement of models on the client, but the default hasn't changed, so if you haven't messed with it then this isn't the problem. That doesn't mean that fiddling with it can't improve the prediction.
I'm more suspicious of the changes in movement code. Sometime before b5 was released, I read on Mantis that the fixes for jetpacks were causing problems with the other forms of movement. While the main problem has certainly been fixed (and thus we have b5 to play with), that doesn't mean there isn't some residual problem.
I have given up using the SG and now go strait to HMG.
Note though that I keep all my settings at default.
It may very well be that messing with your settings like ex_interp and updaterate and cmdrate are screwing you up. It might be best to leave them at defaults. That way you see the same thing that is on everyones screen, and it is how it was meant to be.
Default ex_interp and connection settings.
Then
ex_interp all the way from 0.1 to 0.03
cl_cmdrate from 50-100
cl_updaterate from 50-100
Thanks,
Emanon <!--QuoteEnd--></td></tr></table><div class='postcolor'><!--QuoteEEnd-->
I would just like to remind you.
Forlorn, did you adjust any settings or just get used to where you need too shoot?