<!--quoteo(post=1901829:date=Feb 10 2012, 06:27 PM:name=Max)--><div class='quotetop'>QUOTE (Max @ Feb 10 2012, 06:27 PM) <a href="index.php?act=findpost&pid=1901829"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Those lines that say Client* happened during client side prediction. You don't want to count those. Do something like this:
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->if not Shared.GetIsRunningPrediction() count= count + 1 end<!--c2--></div><!--ec2--><!--QuoteEnd--></div><!--QuoteEEnd--> I guess there's no way to make it work so that you'd see the hits that count on the server, when you're not running the server yourself? Only seeing the client prediction hits is really depressing as each skulk eats up 30+ bullets but doesn't die. Sometimes they also eat 0 shotgun pellets and they do die, which is weird.
Edit: <!--quoteo(post=1902858:date=Feb 14 2012, 06:56 AM:name=Harimau)--><div class='quotetop'>QUOTE (Harimau @ Feb 14 2012, 06:56 AM) <a href="index.php?act=findpost&pid=1902858"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->As someone stated in response to me, that would be limited by the latency, so you'd get the hit feedback 150~250 ms after it hit.<!--QuoteEnd--></div><!--QuoteEEnd--> That wouldn't matter to me. I'd just like to see/hear all my hits so I can justfully blame the game and not my aiming skill. :D
It's just times like these that really ruin the game for me. The fourth shotgun shot really, really should kill the skulk, it's a 100% complete perfect hit but it does nothing. <center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/-CTWXx_sdb4"></param><embed src="http://www.youtube.com/v/-CTWXx_sdb4" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center>
I think 4 shotgun blasts directly into a fade should kill him as well. I had Damage ###### LV.3, 4 shotgun blasts into a fade is REALLY hard to do when he's warping all around, and then all he does is swipe me once.
<!--quoteo(post=1902867:date=Feb 14 2012, 01:20 PM:name=Skie)--><div class='quotetop'>QUOTE (Skie @ Feb 14 2012, 01:20 PM) <a href="index.php?act=findpost&pid=1902867"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->That wouldn't matter to me. I'd just like to see/hear all my hits so I can justfully blame the game and not my aiming skill. :D<!--QuoteEnd--></div><!--QuoteEEnd--> I agree. It wouldn't matter to me either. But it matters to some people. I'd make it an option (client-side prediction for hit feedback, or server-side response for hit feedback). The first looks and sounds "right", but the second is better for getting true hit feedback.
<!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->I guess there's no way to make it work so that you'd see the hits that count on the server, when you're not running the server yourself? Only seeing the client prediction hits is really depressing as each skulk eats up 30+ bullets but doesn't die. Sometimes they also eat 0 shotgun pellets and they do die, which is weird.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't know how it works exactly, but I disabled completely the bullet fire code client side and you still see your hits but delayed by the lag, so I guess the server send you the hits somewhere and you should be able to count them. With a decent ping (<100ms) having the delayed feedback is not a problem, it feels pretty good.
Okay Okay, Yuuki, yeah I read the thread befor I posted lol, I'm just still learning programming so truly I have nothing to offer other than 30 odd years of gaming experience.
It just *feels* very random.... gods I'd never hope for 10 bullets to kill a skulk, man the game would have SOOOOO much a different feel from sept 16th whenever back when....
I was *just asking lol*
Plz note no one has posted suggested clip as of yet too.....I don't think any of us really want that maths to play out....
Well I also feel sometimes like my hits don't register, but after recording some game and watching them in slow motion I couldn't find any evidence of that, so I wonder if the problem is not that my aims sucks. But maybe I was "unlucky" and was on a particular server where everything was working well, I'll try some more latter.
The lack of net_stats is also a problem, do we expect a server with bad tickrate and choke to register hits correctly ?
matsoMaster of PatchesJoin Date: 2002-11-05Member: 7000Members, Forum Moderators, NS2 Developer, Constellation, NS2 Playtester, Squad Five Blue, Squad Five Silver, Squad Five Gold, Reinforced - Shadow, NS2 Community Developer
<!--quoteo(post=1902884:date=Feb 14 2012, 07:20 AM:name=Skie)--><div class='quotetop'>QUOTE (Skie @ Feb 14 2012, 07:20 AM) <a href="index.php?act=findpost&pid=1902884"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->It's just times like these that really ruin the game for me. The fourth shotgun shot really, really should kill the skulk, it's a 100% complete perfect hit but it does nothing.<!--QuoteEnd--></div><!--QuoteEEnd-->
Actually, if you take a careful look at the fourth shot, you can see that there is quite a bit of ground hits off the right side of the skulk ... hard to say exactly how much of course, but probably enough to make it at least plausible that a carapaced skulk would have lived through it - you need 8 lvl 0 shotgun pellets to kill a carap skulk.
<!--quoteo(post=1903117:date=Feb 15 2012, 06:42 AM:name=matso)--><div class='quotetop'>QUOTE (matso @ Feb 15 2012, 06:42 AM) <a href="index.php?act=findpost&pid=1903117"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Actually, if you take a careful look at the fourth shot, you can see that there is quite a bit of ground hits off the right side of the skulk ... hard to say exactly how much of course, but probably enough to make it at least plausible that a carapaced skulk would have lived through it - you need 8 lvl 0 shotgun pellets to kill a carap skulk.<!--QuoteEnd--></div><!--QuoteEEnd--> I'm no sure. not when you take into account the first two shots 'hitting'.
<!--quoteo(post=1903117:date=Feb 14 2012, 07:42 PM:name=matso)--><div class='quotetop'>QUOTE (matso @ Feb 14 2012, 07:42 PM) <a href="index.php?act=findpost&pid=1903117"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Actually, if you take a careful look at the fourth shot, you can see that there is quite a bit of ground hits off the right side of the skulk ... hard to say exactly how much of course, but probably enough to make it at least plausible that a carapaced skulk would have lived through it - you need 8 lvl 0 shotgun pellets to kill a carap skulk.<!--QuoteEnd--></div><!--QuoteEEnd-->
The skulk (if we believe the top right notification) died just after 16 bullets, but I probably missed a few there. Looking at it again it seems like the death animation is delayed or something, I still fire ten bullets even though I had the top right notification.
Also, the kill notification for the drifter is displayed when it still have 30% health.
<!--quoteo(post=1902884:date=Feb 14 2012, 02:20 AM:name=Skie)--><div class='quotetop'>QUOTE (Skie @ Feb 14 2012, 02:20 AM) <a href="index.php?act=findpost&pid=1902884"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->It's just times like these that really ruin the game for me. The fourth shotgun shot really, really should kill the skulk, it's a 100% complete perfect hit but it does nothing. <center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/-CTWXx_sdb4"></param><embed src="http://www.youtube.com/v/-CTWXx_sdb4" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center><!--QuoteEnd--></div><!--QuoteEEnd-->
Holy crap did you see that latency? You were half-cocked and THEN the skulk died.
Also it does feel like the game's frame rate is important. I noticed that my LMG is considerably less effective, and I have to aim further in front, in low fps situations.
<!--quoteo(post=1912753:date=Mar 14 2012, 04:48 AM:name=Anticept)--><div class='quotetop'>QUOTE (Anticept @ Mar 14 2012, 04:48 AM) <a href="index.php?act=findpost&pid=1912753"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Holy crap did you see that latency? You were half-cocked and THEN the skulk died.
Also it does feel like the game's frame rate is important. I noticed that my LMG is considerably less effective, and I have to aim further in front, in low fps situations.<!--QuoteEnd--></div><!--QuoteEEnd-->
Frame rate determines your rate of fire.
This rapid fire hit registration thing may be imaginary. Its very easy in network games to think things are taking far more bullets than they should. Counter-strike is a prime example, considering it only takes 3 bullets from most weapons to kill someone. First thing is, first shots are very rarely on target with automatics, so you can write off 3 to 4 bullets there. Second, although its minor, the LMG does have a spread, and even perfect aim will not result in 100% hit rate. Third is the delay on death, unavoidable in all network games. Even once you have hit the skulk with those 10 bullets, its still going to be at least 1 second before the server updates him as dead, which is another 5-10 bullets wasted shooting at a target that is already down. So in the 10 that you fire to kill, I think it would be reasonable to say that 15 are not going to hit, and are easy to assume were "lost".
Easy to test though, join a fullish server with a friend, find a quiet corner, turn afterburner recording on at max FPS and empty a clip into a stationery skulk. Replay it in slow motion and watch the bullet counter. Anything under 15 bullets is perfectly acceptable I would think. This would show that the bullets are not being lost, as the target does not need to be moving to test this.
<!--quoteo(post=1912836:date=Mar 14 2012, 11:15 AM:name=Imbalanxd)--><div class='quotetop'>QUOTE (Imbalanxd @ Mar 14 2012, 11:15 AM) <a href="index.php?act=findpost&pid=1912836"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Third is the delay on death, unavoidable in all network games. Even once you have hit the skulk with those 10 bullets, its still going to be at least 1 second before the server updates him as dead, which is another 5-10 bullets wasted shooting at a target that is already down. So in the 10 that you fire to kill, I think it would be reasonable to say that 15 are not going to hit, and are easy to assume were "lost".<!--QuoteEnd--></div><!--QuoteEEnd--> This can't be true. In no other multiplayer FPS game that I've played has it taken 1 second before hits register and I see the player as dead / taking damage. In ns2 you are receiving packets from the server every 50ms, so it shouldn't take much more than that if you are playing on a low ping server. There is something wrong that is causing a massive delay and makes it feel like your constantly playing on a 150+ ping server. It even affects the game when you are playing on a lan.
<!--quoteo(post=1912841:date=Mar 14 2012, 01:34 PM:name=Wilson)--><div class='quotetop'>QUOTE (Wilson @ Mar 14 2012, 01:34 PM) <a href="index.php?act=findpost&pid=1912841"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->This can't be true. In no other multiplayer FPS game that I've played has it taken 1 second before hits register and I see the player as dead / taking damage. In ns2 you are receiving packets from the server every 50ms, so it shouldn't take much more than that if you are playing on a low ping server. There is something wrong that is causing a massive delay and makes it feel like your constantly playing on a 150+ ping server. It even affects the game when you are playing on a lan.<!--QuoteEnd--></div><!--QuoteEEnd-->
Both source and frostbite 2 do it. It can't be ping alone that causes it, otherwise it would simply be a case of a 50-100ms delay. I think it has more to do with synchronising the death event between clients. I think every game I've ever played that wasn't incredibly simple had this "problem". And no its not my connection, this happens to everyone, but the game needs to have a shotgun or similar mechanic in order to recognise it.
Delay on death doesn't happen for me in source. When I kill a guy the delay is only about 50ms or whatever my ping is. Not anywhere near 1 second. Even in ns2 the dealy isn't 1 second, it's about 150-250ms and that is too long imo.
<!--quoteo(post=1902376:date=Feb 12 2012, 12:17 PM:name=Yuuki)--><div class='quotetop'>QUOTE (Yuuki @ Feb 12 2012, 12:17 PM) <a href="index.php?act=findpost&pid=1902376"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->If I understand correctly what max said, this is impossible.<!--QuoteEnd--></div><!--QuoteEEnd-->
Max's claim is what seems impossible to me. Latency varies, clocks are not necessarily consistent(E.g. QueryPerformanceCounter has had bugs in the past on certain hardware and presumably still does; this may result in say, a few ms jitter).
I see no way to perfectly reconstruct what the client saw, even if no bugs are present.
<!--quoteo(post=1912866:date=Mar 14 2012, 01:58 PM:name=ogz)--><div class='quotetop'>QUOTE (ogz @ Mar 14 2012, 01:58 PM) <a href="index.php?act=findpost&pid=1912866"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->how do i turn off client hit prediction? i wana see what the server tells me so I know whats happening for real
then i can learn to play with lag, just like good ole quake days<!--QuoteEnd--></div><!--QuoteEEnd-->
AFAIK, you can't. All the net settings are locked down for now.
<!--quoteo(post=1903148:date=Feb 14 2012, 03:21 PM:name=Yuuki)--><div class='quotetop'>QUOTE (Yuuki @ Feb 14 2012, 03:21 PM) <a href="index.php?act=findpost&pid=1903148"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Also, the kill notification for the drifter is displayed when it still have 30% health.<!--QuoteEnd--></div><!--QuoteEEnd-->
// Get information about entity when looking at it. function GetCrosshairText(entity, teamNumber) . . . // But aliens only ever need total % text = string.format("%s (%d%%)", text, math.ceil(entity:GetHealthScalar()*100)) . . .<!--c2--></div><!--ec2-->
That function is called by:
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->function Player:UpdateCrossHairText() -> function Player:UpdateMisc(input) ([b]in Player_Client.lua[/b])
there's a tons of calls before on client side, but "if not Shared.GetIsRunningPrediction() then" might avoid the problem that you can see in the video. If I understand correctly this function only executes when the client have good data from the server (aka real data), maybe they added this line from your post to now, I dont see this part of code until now.
However I have a lot of problems understanding how client and server works. Sometimes you see five, six calls from one function in client-side, one in server-side -prediction i know-. Sometimes I wrote some code that made changes in entities -e.g. origin related-, I can see the debug output from server (and client too, four or five times), but the changes dont affect the world. Only in some random time for me, the origin is modified correctly. I know it, its a problem with my code but I find it hard to understand.
<!--quoteo(post=1902860:date=Feb 13 2012, 11:58 PM:name=Krizzen)--><div class='quotetop'>QUOTE (Krizzen @ Feb 13 2012, 11:58 PM) <a href="index.php?act=findpost&pid=1902860"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Might as well go back to a Quake style lag system (for the uninitiated: aim ahead. Way ahead) so we can b**ch about LPBs schooling us.<!--QuoteEnd--></div><!--QuoteEEnd--> This is fine with me. We are all LPBs nowadays.
I definitely feel something is wrong with the hit reg atm. At first I only thought it was my aim but after playing for awhile I just sometime know a skulk should be a dead when he isnt. I feel this is more apperent as my fps drop or if my latency is higher. if a skulks run straight at me I can usally kill him in 10 shots, same with skulks who are standing still. However when they start doing basic manuveres I start getting the feeling I'm shooting a wall, I can see plenty of blood and my corsair is on target but still it somewhat survives, sometimes with 35+ hp(know since they dont die when I hit them with my gun).
I know of course that ill waste extra bullets when latency is higher and when the skulks is not running in a straight line, but from experience I just know there is something wrong with the reg. Played ns1 for years, ns1 was not the best game with hitreg but I rarly got this feeling there.
I think this has something to do with prediction, does anyone remember the command to turn off prediction so you can see the game from server side(with the delay). Would be nice for testing.
I'm also wondering if we see this problem on low ping servers, a test with 2 or more machines on lan, I might be able to do that.
I think a large part of it is to do with this: <a href="http://www.unknownworlds.com/ns2/forums/index.php?showtopic=116894&st=0&start=0" target="_blank">http://www.unknownworlds.com/ns2/forums/in...t=0&start=0</a>
There is a 150-250ms delay even when on a lan server. That's enough for you to fire several more bullets than needed into skulks.
<!--quoteo(post=1912899:date=Mar 14 2012, 05:48 PM:name=Wilson)--><div class='quotetop'>QUOTE (Wilson @ Mar 14 2012, 05:48 PM) <a href="index.php?act=findpost&pid=1912899"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I think a large part of it is to do with this: <a href="http://www.unknownworlds.com/ns2/forums/index.php?showtopic=116894&st=0&start=0" target="_blank">http://www.unknownworlds.com/ns2/forums/in...t=0&start=0</a>
There is a 150-250ms delay even when on a lan server. That's enough for you to fire several more bullets than needed into skulks.<!--QuoteEnd--></div><!--QuoteEEnd-->
Where is this delay coming from? Especial on a lan server it should not be there. Only a server tick rate of 4-5 would normally cause this. I encountered a similar problem at work with the nagle algorithm. But this is only TCP related and the game communication should be on UDP.
I think it's the 100ms interpolation plus ping. Not sure though.
But I did the same thing than skie with the command center and it was pretty much perfect, except the hit animation lagging behind (max said there was a bug with it).
<!--quoteo(post=1912904:date=Mar 14 2012, 10:12 AM:name=Yuuki)--><div class='quotetop'>QUOTE (Yuuki @ Mar 14 2012, 10:12 AM) <a href="index.php?act=findpost&pid=1912904"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I think it's the 100ms interpolation plus ping. Not sure though.
But I did the same thing than skie with the command center and it was pretty much perfect, except the hit animation lagging behind (max said there was a bug with it).<!--QuoteEnd--></div><!--QuoteEEnd--> The game uses 150ms interpolation at the moment (will be reduced to 100ms when we ship).
<!--quoteo(post=1912932:date=Mar 14 2012, 01:50 PM:name=Max)--><div class='quotetop'>QUOTE (Max @ Mar 14 2012, 01:50 PM) <a href="index.php?act=findpost&pid=1912932"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->The game uses 150ms interpolation at the moment (will be reduced to 100ms when we ship).<!--QuoteEnd--></div><!--QuoteEEnd--> Is there anyway to determine interp in real time based the latency of those entities involved?
swalkSay hello to my little friend.Join Date: 2011-01-20Member: 78384Members, Squad Five Blue
<!--quoteo(post=1912935:date=Mar 14 2012, 07:59 PM:name=Kalabalana)--><div class='quotetop'>QUOTE (Kalabalana @ Mar 14 2012, 07:59 PM) <a href="index.php?act=findpost&pid=1912935"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Is there anyway to determine interp in real time based the latency of those entities involved?<!--QuoteEnd--></div><!--QuoteEEnd--> I had the same thought. That would make the interp fit with all latencies. Either that, or make it a manual setting, so we can get better interpolation on lowping/higherping servers. It's not a good idea to assume that everyone will be playing at 150ms or 100ms.
<!--quoteo(post=1912937:date=Mar 14 2012, 02:06 PM:name=swalk)--><div class='quotetop'>QUOTE (swalk @ Mar 14 2012, 02:06 PM) <a href="index.php?act=findpost&pid=1912937"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I had the same thought. That would make the interp fit with all latencies. Either that, or make it a manual setting, so we can get better interpolation on lowping/higherping servers. It's not a good idea to assume that everyone will be playing at 150ms or 100ms.<!--QuoteEnd--></div><!--QuoteEEnd--> Yeah. Otherwise I'm thinking of how much tweaking I had to do with my interp back in cs.
Comments
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->if not Shared.GetIsRunningPrediction()
count= count + 1
end<!--c2--></div><!--ec2--><!--QuoteEnd--></div><!--QuoteEEnd-->
I guess there's no way to make it work so that you'd see the hits that count on the server, when you're not running the server yourself?
Only seeing the client prediction hits is really depressing as each skulk eats up 30+ bullets but doesn't die.
Sometimes they also eat 0 shotgun pellets and they do die, which is weird.
Edit:
<!--quoteo(post=1902858:date=Feb 14 2012, 06:56 AM:name=Harimau)--><div class='quotetop'>QUOTE (Harimau @ Feb 14 2012, 06:56 AM) <a href="index.php?act=findpost&pid=1902858"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->As someone stated in response to me, that would be limited by the latency, so you'd get the hit feedback 150~250 ms after it hit.<!--QuoteEnd--></div><!--QuoteEEnd-->
That wouldn't matter to me. I'd just like to see/hear all my hits so I can justfully blame the game and not my aiming skill. :D
<center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/-CTWXx_sdb4"></param><embed src="http://www.youtube.com/v/-CTWXx_sdb4" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center>
I think 4 shotgun blasts directly into a fade should kill him as well. I had Damage ###### LV.3, 4 shotgun blasts into a fade is REALLY hard to do when he's warping all around, and then all he does is swipe me once.
He doesn't even have to aim.
Pisses me the hell off.
I agree. It wouldn't matter to me either. But it matters to some people. I'd make it an option (client-side prediction for hit feedback, or server-side response for hit feedback). The first looks and sounds "right", but the second is better for getting true hit feedback.
Only seeing the client prediction hits is really depressing as each skulk eats up 30+ bullets but doesn't die.
Sometimes they also eat 0 shotgun pellets and they do die, which is weird.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't know how it works exactly, but I disabled completely the bullet fire code client side and you still see your hits but delayed by the lag, so I guess the server send you the hits somewhere and you should be able to count them.
With a decent ping (<100ms) having the delayed feedback is not a problem, it feels pretty good.
It just *feels* very random.... gods I'd never hope for 10 bullets to kill a skulk, man the game would have SOOOOO much a different feel from sept 16th whenever back when....
I was *just asking lol*
Plz note no one has posted suggested clip as of yet too.....I don't think any of us really want that maths to play out....
The lack of net_stats is also a problem, do we expect a server with bad tickrate and choke to register hits correctly ?
Actually, if you take a careful look at the fourth shot, you can see that there is quite a bit of ground hits off the right side of the skulk ... hard to say exactly how much of course, but probably enough to make it at least plausible that a carapaced skulk would have lived through it - you need 8 lvl 0 shotgun pellets to kill a carap skulk.
I'm no sure. not when you take into account the first two shots 'hitting'.
If you noticed, he had lvl 2 upgrades
I did some more recordings, but didn't found much. I had a good ping but the server was a bit choppy :
<center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/y_1SgG2CI4s"></param><embed src="http://www.youtube.com/v/y_1SgG2CI4s" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center>
The skulk (if we believe the top right notification) died just after 16 bullets, but I probably missed a few there. Looking at it again it seems like the death animation is delayed or something, I still fire ten bullets even though I had the top right notification.
Also, the kill notification for the drifter is displayed when it still have 30% health.
<center><object width="450" height="356"><param name="movie" value="http://www.youtube.com/v/-CTWXx_sdb4"></param><embed src="http://www.youtube.com/v/-CTWXx_sdb4" type="application/x-shockwave-flash" width="450" height="356"></embed></object></center><!--QuoteEnd--></div><!--QuoteEEnd-->
Holy crap did you see that latency? You were half-cocked and THEN the skulk died.
Also it does feel like the game's frame rate is important. I noticed that my LMG is considerably less effective, and I have to aim further in front, in low fps situations.
Also it does feel like the game's frame rate is important. I noticed that my LMG is considerably less effective, and I have to aim further in front, in low fps situations.<!--QuoteEnd--></div><!--QuoteEEnd-->
Frame rate determines your rate of fire.
This rapid fire hit registration thing may be imaginary. Its very easy in network games to think things are taking far more bullets than they should. Counter-strike is a prime example, considering it only takes 3 bullets from most weapons to kill someone. First thing is, first shots are very rarely on target with automatics, so you can write off 3 to 4 bullets there. Second, although its minor, the LMG does have a spread, and even perfect aim will not result in 100% hit rate. Third is the delay on death, unavoidable in all network games. Even once you have hit the skulk with those 10 bullets, its still going to be at least 1 second before the server updates him as dead, which is another 5-10 bullets wasted shooting at a target that is already down. So in the 10 that you fire to kill, I think it would be reasonable to say that 15 are not going to hit, and are easy to assume were "lost".
Easy to test though, join a fullish server with a friend, find a quiet corner, turn afterburner recording on at max FPS and empty a clip into a stationery skulk. Replay it in slow motion and watch the bullet counter. Anything under 15 bullets is perfectly acceptable I would think. This would show that the bullets are not being lost, as the target does not need to be moving to test this.
This can't be true. In no other multiplayer FPS game that I've played has it taken 1 second before hits register and I see the player as dead / taking damage. In ns2 you are receiving packets from the server every 50ms, so it shouldn't take much more than that if you are playing on a low ping server. There is something wrong that is causing a massive delay and makes it feel like your constantly playing on a 150+ ping server. It even affects the game when you are playing on a lan.
Both source and frostbite 2 do it. It can't be ping alone that causes it, otherwise it would simply be a case of a 50-100ms delay. I think it has more to do with synchronising the death event between clients. I think every game I've ever played that wasn't incredibly simple had this "problem". And no its not my connection, this happens to everyone, but the game needs to have a shotgun or similar mechanic in order to recognise it.
Max's claim is what seems impossible to me. Latency varies, clocks are not necessarily consistent(E.g. QueryPerformanceCounter has had bugs in the past on certain hardware and presumably still does; this may result in say, a few ms jitter).
I see no way to perfectly reconstruct what the client saw, even if no bugs are present.
then i can learn to play with lag, just like good ole quake days
then i can learn to play with lag, just like good ole quake days<!--QuoteEnd--></div><!--QuoteEEnd-->
AFAIK, you can't. All the net settings are locked down for now.
<!--quoteo(post=1903148:date=Feb 14 2012, 03:21 PM:name=Yuuki)--><div class='quotetop'>QUOTE (Yuuki @ Feb 14 2012, 03:21 PM) <a href="index.php?act=findpost&pid=1903148"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Also, the kill notification for the drifter is displayed when it still have 30% health.<!--QuoteEnd--></div><!--QuoteEEnd-->
I CAN HELP HERE!
I'm coding yesterday and find some weirds things.
Some random parts of code:
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->NS2Utility.lua
// Get information about entity when looking at it.
function GetCrosshairText(entity, teamNumber)
.
.
.
// But aliens only ever need total %
text = string.format("%s (%d%%)", text, math.ceil(entity:GetHealthScalar()*100))
.
.
.<!--c2--></div><!--ec2-->
That function is called by:
<!--c1--><div class='codetop'>CODE</div><div class='codemain'><!--ec1-->function Player:UpdateCrossHairText() -> function Player:UpdateMisc(input) ([b]in Player_Client.lua[/b])
PROFILE("Player:UpdateMisc")
if not Shared.GetIsRunningPrediction() then
self:UpdateCrossHairText()
self:UpdateDamageIndicators()
self:UpdateChat(input)
end
end<!--c2--></div><!--ec2-->
there's a tons of calls before on client side, but "if not Shared.GetIsRunningPrediction() then" might avoid the problem that you can see in the video. If I understand correctly this function only executes when the client have good data from the server (aka real data), maybe they added this line from your post to now, I dont see this part of code until now.
However I have a lot of problems understanding how client and server works. Sometimes you see five, six calls from one function in client-side, one in server-side -prediction i know-. Sometimes I wrote some code that made changes in entities -e.g. origin related-, I can see the debug output from server (and client too, four or five times), but the changes dont affect the world. Only in some random time for me, the origin is modified correctly. I know it, its a problem with my code but I find it hard to understand.
This is fine with me. We are all LPBs nowadays.
I feel this is more apperent as my fps drop or if my latency is higher. if a skulks run straight at me I can usally kill him in 10 shots, same with skulks who are standing still. However when they start doing basic manuveres I start getting the feeling I'm shooting a wall, I can see plenty of blood and my corsair is on target but still it somewhat survives, sometimes with 35+ hp(know since they dont die when I hit them with my gun).
I know of course that ill waste extra bullets when latency is higher and when the skulks is not running in a straight line, but from experience I just know there is something wrong with the reg. Played ns1 for years, ns1 was not the best game with hitreg but I rarly got this feeling there.
I think this has something to do with prediction, does anyone remember the command to turn off prediction so you can see the game from server side(with the delay). Would be nice for testing.
I'm also wondering if we see this problem on low ping servers, a test with 2 or more machines on lan, I might be able to do that.
There is a 150-250ms delay even when on a lan server. That's enough for you to fire several more bullets than needed into skulks.
There is a 150-250ms delay even when on a lan server. That's enough for you to fire several more bullets than needed into skulks.<!--QuoteEnd--></div><!--QuoteEEnd-->
Where is this delay coming from? Especial on a lan server it should not be there. Only a server tick rate of 4-5 would normally cause this.
I encountered a similar problem at work with the nagle algorithm. But this is only TCP related and the game communication should be on UDP.
But I did the same thing than skie with the command center and it was pretty much perfect, except the hit animation lagging behind (max said there was a bug with it).
But I did the same thing than skie with the command center and it was pretty much perfect, except the hit animation lagging behind (max said there was a bug with it).<!--QuoteEnd--></div><!--QuoteEEnd-->
The game uses 150ms interpolation at the moment (will be reduced to 100ms when we ship).
Is there anyway to determine interp in real time based the latency of those entities involved?
I had the same thought.
That would make the interp fit with all latencies.
Either that, or make it a manual setting, so we can get better interpolation on lowping/higherping servers.
It's not a good idea to assume that everyone will be playing at 150ms or 100ms.
That would make the interp fit with all latencies.
Either that, or make it a manual setting, so we can get better interpolation on lowping/higherping servers.
It's not a good idea to assume that everyone will be playing at 150ms or 100ms.<!--QuoteEnd--></div><!--QuoteEEnd-->
Yeah. Otherwise I'm thinking of how much tweaking I had to do with my interp back in cs.