Hit Detection

Comments

  • Rev0verDrive
    6050 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    oJU5T1No wrote: »
    I experienced alot of de-sync and lag and other strange latency related things pre server side hit dection and I don't think the server side hit detection has impacted any of this, It was suspected pre-patch high latency players with unstable connections were lagging out the servers now with ping in the scoreboard theres some evidence to support this.

    Well you shouldn't really "lag" because of hp's do to the netcode and Lag comp changes. Desync should've been resolved.
  • misisipiRivrRat
    451 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    Most of the tdm servers I've been playing in have at least 3 players over 150ms, (this is up since the hotfix) and hit detection seems to to be hit or miss, pardon the pun. When the patch rolled out I saw several high latency players to begin with but they thinned out fairly fast but since the hot patch they are back. Now almost every game I'm in has pre-patch stuff going on, I can't prove it's because of high pingers but it seems like more than coincidence to me.

    Just my 2 cents.

    Hi John,
    Just played a few matches of Frontlines and they were terrible . This is so friggin' frustrating right now. Finally had enough when a enemy sniper ran by me and I started unloading on him. His character was being hit but no hit markers. Turns and kills me. By the way he had a low ping. @vball. Can I ask which us region you play, east or west? You say you aren't seeing many high pingers. It's the opposite for me. 5-10 every match of Frontlines. Maybe that isn't a lot? I mostly play us east. Also I saw a few guys with very low ping maybe like what you've seen in your netgraph. To guys on the enemy team had pings of 11ms as did a guy on my team.
  • mmarkweII
    2887 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    edited May 10
    @Rev0verDrive In your opinion, do you think it would be beneficial to run 3 instances (instead of 4) on a server to save resources for that server if/when needed?

    @misisipiRivrRat we usually play US East mostly between 9 til midnight eastern (sometimes later). CQL. Smaller game modes should be at 60hz now, so any issues on those servers would be noticed easier, correct?
  • misisipiRivrRat
    451 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member

    @vballmvp,
    The reason I asked where you play. Look at this short vid of some different servers I got into. There are way more than a couple of high ping players. These are east and west us servers recorded on a weekend between 5 and 6am central.
  • Rev0verDrive
    6050 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    mmarkweII wrote: »
    @Rev0verDrive In your opinion, do you think it would be beneficial to run 3 instances (instead of 4) on a server to save resources for that server if/when needed?

    Subjectively yeah, why not? just split the resources evenly over 3 instances. Objectively I couldn't say without knowing certain aspects of the code structure and what the resources are and if there are allocations. optimally you want all 4 instances running efficiently. Cost issue. If they had to drop to 3, how would that impact rental costs?
  • denjoga
    607 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1 Member
    edited May 10
    It's a known bug, because the devs know about it, me saying it's known, doesn't change any of that. They've publicly asked for help in finding the issue, that's how I know.

    It might've been the same bug pre-patch, might be the hit reg bug from bf4. It could be totally new. You don't know.

    Sorry, but there's no way that zero hit bug existed pre-patch.
    It is so severe - hits show in player animation but register no markers or damage - and consistent (when it happens - it's either there constantly or not at all, requiring quit and restart to resolve) that there would certainly have been a noticeable outcry in the forums and elsewhere if it had existed before the patch.
    I believe (but am too lazy to search it out; feel free to correct me if I'm wrong) that Mischkag stated that it was a bug that was introduced with the patch.
    If you look at the videos of the bug in action, it is so severe and blatant that there's just no way it would have gone unnoticed/unmentioned if it existed before the patch.

    (I don't know anything about a BF4 hit reg bug.)
  • KingTolapsium
    5482 postsMember, Battlefield 3, Battlefield 4, Battlefield Hardline, Battlefield, Battlefield 1, CTE, BF1IncursionsAlpha Member
    denjoga wrote: »
    It's a known bug, because the devs know about it, me saying it's known, doesn't change any of that. They've publicly asked for help in finding the issue, that's how I know.

    It might've been the same bug pre-patch, might be the hit reg bug from bf4. It could be totally new. You don't know.

    Sorry, but there's no way that zero hit bug existed pre-patch.
    It is so severe - hits show in player animation but register no markers or damage - and consistent (when it happens - it's either there constantly or not at all, requiring quit and restart to resolve) that there would certainly have been a noticeable outcry in the forums and elsewhere if it had existed before the patch.
    I believe (but am too lazy to search it out; feel free to correct me if I'm wrong) that Mischkag stated that it was a bug that was introduced with the patch.
    If you look at the videos of the bug in action, it is so severe and blatant that there's just no way it would have gone unnoticed/unmentioned if it existed before the patch.

    (I don't know anything about a BF4 hit reg bug.)

    The bf4 "bullet sponge" bug was almost visually identical. Same thing could happen in bf1, but very rarely, I don't think I've seen it since the beta? Well before the patch anyways.

    It's possible reducing the frame history exposed this previously hidden bug.

    All I know is it exists, tracing it's origin doesn't really benefit us much. I'm more concerned with it getting fixed.
  • VBALL_MVP
    4099 postsMember, Battlefield 3, Battlefield 4, Battlefield Hardline, Battlefield, Battlefield 1 Member
    edited May 11

    @vballmvp,
    The reason I asked where you play. Look at this short vid of some different servers I got into. There are way more than a couple of high ping players. These are east and west us servers recorded on a weekend between 5 and 6am central.

    As @mmarkweII said I play US East, roughly around 10pm CST. I dont like playing above 60ms (doesnt feel smooth to me) so i avoid west. Usually see 2 per team 3 max....when playing a full server that's not much. Also I am basing that on many other people's comments across multiple threads.

    Not saying there are not instance that people have had more, as you illustrated, but the question is did it feel worse in those instances and what was Srvtick at those times.
  • KingTolapsium
    5482 postsMember, Battlefield 3, Battlefield 4, Battlefield Hardline, Battlefield, Battlefield 1, CTE, BF1IncursionsAlpha Member
    @mischkag Out of compromise we get the best solutions.

    Possibly the most interesting post you've thrown onto the forums yet! :)

    It's unfortunate that the changes required and the current player pools won't allow dual official server sets in the short term. It sounds like the future will be quite bright in this regard.

    Hopefully the private server browsing experience can be improved to highlight private servers with a hard ping cap. Not your department obviously.

    Has the team been able to address the hit detection bug??? Or are instances of occurrence at least down on CTE?
  • kniphtee
    244 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    @mischkag - if I understood 51% of what you just deliberated it would be quarter of what you know about your trade - bravo. One humble question, is there a sync issue when joining a game late? Many times I join a game halfway through and everything is off, or glitch, or way out of balance and then next map and the following maps all things are honkydory (and this is not due to rust first game back in the saddle as I play on and off when I get time - it is simply first game joining already in progress regardless of how many times I've jumped on that day)? I am amazed at the work put in and the effort put forth with hit reg being way better than previously. Thank you much - keep it up.
  • mmarkweII
    2887 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    mischkag wrote: »
    About the snagging. I explained it before. The higher the ping, the higher the divergence of your game, the servers and the other clients. There are some shortcomings in how it tries to solve the physics there, but the bottom line is that this is just another example of how high ping players create issues. Physics does not give a low ping player some priority (i know you would love that :)).

    As mentioned before, region locks would split the player base further and require changes on the FrontEnd UI, Matchmaking, Server Browser, Operations infrastructure to just name a few.
    We need to first aim for rental servers to get a ping limit. Thats in progress.
    About the OOR friends only invites, as long as you have the server browser showing all servers and you are able to join anywhere, we cannot move forward with anything like that.

    We all know that the solution is to have different sets of servers with region locks etc, but i personally believe that we wont get there at this point for BF1.
    Too many regions have already empty servers. So it would only make sense for the most populated ones. It is infact a rather big change to handle that particularly as the actual server centers do not coincide with the QOS site which you use ICMP requests to before joining. So it is perfectly possible to either have a good connection to the Quality of service point with ICMP and than a rather poor connection to the actual server (less of a problem in US, more in Europe and elsewhere) or vice versa.
    We learned from that and the necessary infrastructure for upcoming games will use a new system and infrastructure. It may come for BF1 later this year, but i would rather be very pessimistic about it as it involves too many unknowns and manpower to get this right.
    I hope that once the server side hit detection works, that players will go back more to their own regions instead of creating all these inherent high ping issues.

    Delayed damage application can only be solved thru the server side hit detection or ping caps. As the latter will never come for this game or late, i pushed to get at least something done.
    The snagging can most likely be only solved by ping caps.
    In the meantime i will keep pushing to minimize effects of high ping jitter as the biggest issue in my opinion is shooting at teleporting ghosts.

    We assisted Chris on his great video. I do agree that Chris's video should be a mandatory watch, but i am just an engineer and the best bet is to make sure thru social channels you are all part of, that it gets shared and watched :)

    The next patch will hopefully see the Server side hit registration working flawlessly, people get used to it and then we gradually may lower it to improve the experience. This forum is drastically different with mostly low ping players whereas reddit and Twitter seems more used by out of region players(at least when they get affected) and that all makes it hard to fully get the real picture and what a good compromise looks like.
    On a personal note, i play on US West. This is by far the location out of the popular ones, with the least percentage of high ping players (i can also tell from some proper telemetry data i got). The ping from South Africa to US East is much better than to Europe for some ISP reasons. US East is a very popular spot to be joined by South America, South Africa and even Europe. So i have to play elsewhere to not be spoiled by what we mostly have on US West (not saying that it is all great here). So as much as i agree with most what you say here, i have to also listen to players from other regions and their struggles to find a global or at least per region compromise.

    With regards to server costs and server side hit registration. It is true on paper that it costs more performance. But the extra costs only kick in the moment a bullet by a high ping player actually passes the arbitration. So i could not measure any meaningful difference even with all 64 players running thru this code path during development.

    I did do a change to the servers socket handling with regards to server load. It may be the reason why u always thought that even on a 30hz server it felt like it degraded after seeing server ticks >20ms. But it is hard to prove in a development environment because you need to essentially fill a cloud box with multiple server instances to get this happening so that multiple processes and threads interact with each other.
    Hope for some good feedback with the next update (and no, i am not at liberty to say when that is).

    We had playtests where we artificially host far away so that all 64 players have pings >190ms. Still, no server degrade could be measured. Granted, the coorporate network is not equal to some random ISPs with Wifi and buffer bloats, but i still could not see any meaningful performance drops but meaningful ping variations.
    Now it does have an impact in the frame to frame absolute time which subsequently affects the UDP send time difference which eventually results in ping jitter on the receiving end. This leads to more lag compensation etc.
    I will keep looking into it. Of course, packet loss results in more computation costs as you have to mark the lost states again. Upstream packet loss is no problem, but the jitter results in the varying server loads. It should even out, but that highly depends. 60Hz servers have less players, but are also more prone to these variances.

    GREAT post! Thank you for the continued feedback and for providing A LOT of info behind the scenes too.
  • lizzard
    953 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1 Member
    mischkag wrote: »
    About the snagging. I explained it before. The higher the ping, the higher the divergence of your game, the servers and the other clients. There are some shortcomings in how it tries to solve the physics there, but the bottom line is that this is just another example of how high ping players create issues. Physics does not give a low ping player some priority (i know you would love that :)).

    As mentioned before, region locks would split the player base further and require changes on the FrontEnd UI, Matchmaking, Server Browser, Operations infrastructure to just name a few.
    We need to first aim for rental servers to get a ping limit. Thats in progress.
    About the OOR friends only invites, as long as you have the server browser showing all servers and you are able to join anywhere, we cannot move forward with anything like that.

    We all know that the solution is to have different sets of servers with region locks etc, but i personally believe that we wont get there at this point for BF1.
    Too many regions have already empty servers. So it would only make sense for the most populated ones. It is infact a rather big change to handle that particularly as the actual server centers do not coincide with the QOS site which you use ICMP requests to before joining. So it is perfectly possible to either have a good connection to the Quality of service point with ICMP and than a rather poor connection to the actual server (less of a problem in US, more in Europe and elsewhere) or vice versa.
    We learned from that and the necessary infrastructure for upcoming games will use a new system and infrastructure. It may come for BF1 later this year, but i would rather be very pessimistic about it as it involves too many unknowns and manpower to get this right.
    I hope that once the server side hit detection works, that players will go back more to their own regions instead of creating all these inherent high ping issues.

    Delayed damage application can only be solved thru the server side hit detection or ping caps. As the latter will never come for this game or late, i pushed to get at least something done.
    The snagging can most likely be only solved by ping caps.
    In the meantime i will keep pushing to minimize effects of high ping jitter as the biggest issue in my opinion is shooting at teleporting ghosts.

    We assisted Chris on his great video. I do agree that Chris's video should be a mandatory watch, but i am just an engineer and the best bet is to make sure thru social channels you are all part of, that it gets shared and watched :)

    The next patch will hopefully see the Server side hit registration working flawlessly, people get used to it and then we gradually may lower it to improve the experience. This forum is drastically different with mostly low ping players whereas reddit and Twitter seems more used by out of region players(at least when they get affected) and that all makes it hard to fully get the real picture and what a good compromise looks like.
    On a personal note, i play on US West. This is by far the location out of the popular ones, with the least percentage of high ping players (i can also tell from some proper telemetry data i got). The ping from South Africa to US East is much better than to Europe for some ISP reasons. US East is a very popular spot to be joined by South America, South Africa and even Europe. So i have to play elsewhere to not be spoiled by what we mostly have on US West (not saying that it is all great here). So as much as i agree with most what you say here, i have to also listen to players from other regions and their struggles to find a global or at least per region compromise.

    With regards to server costs and server side hit registration. It is true on paper that it costs more performance. But the extra costs only kick in the moment a bullet by a high ping player actually passes the arbitration. So i could not measure any meaningful difference even with all 64 players running thru this code path during development.

    I did do a change to the servers socket handling with regards to server load. It may be the reason why u always thought that even on a 30hz server it felt like it degraded after seeing server ticks >20ms. But it is hard to prove in a development environment because you need to essentially fill a cloud box with multiple server instances to get this happening so that multiple processes and threads interact with each other.
    Hope for some good feedback with the next update (and no, i am not at liberty to say when that is).

    We had playtests where we artificially host far away so that all 64 players have pings >190ms. Still, no server degrade could be measured. Granted, the coorporate network is not equal to some random ISPs with Wifi and buffer bloats, but i still could not see any meaningful performance drops but meaningful ping variations.
    Now it does have an impact in the frame to frame absolute time which subsequently affects the UDP send time difference which eventually results in ping jitter on the receiving end. This leads to more lag compensation etc.
    I will keep looking into it. Of course, packet loss results in more computation costs as you have to mark the lost states again. Upstream packet loss is no problem, but the jitter results in the varying server loads. It should even out, but that highly depends. 60Hz servers have less players, but are also more prone to these variances.

    Nice awnser mischkag.
    A bitt hard to read and clearly understand, for a non American, with limited netcode knowledge maybe. But its better than no information at all, like we're used to since previous titles. :smile:

    To lay it in a simple way.
    You have not seen any server performance impact, after the serverside hitreg implementation.

    You do play the game, and test to play the game with different ping walues.
    (this is great. It shows that you are not only a employee or paid to do the programming, but also a gamer.)

    One thing I cant clearly make out from your post.
    Is if there can be some form of serverdegradation / gameplay impact, without the SrvTick goes up against the upper limit?

    Like many players have had a feeling of since bf4 atleast.

    Everything looks good on the graf, and on different connection tests.
    but the gameplay feels wery inconsistent..

    Most noticeable for me have been following things.

    Sounds go missing or goes in repeat.

    Suppression effects is to low or abnormaly high.

    Many hits on players don't count.

    One game it feels like your behind the entire enemy team, like one or a half second behind the action.
    The next round it feels to easy, and its like emthe enemy team is behind instead.

    What can be the case of the above?

    Thanks again for putting up with the uneducated questions :smile:
  • denjoga
    607 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1 Member
    edited May 11
    Chris just posted his analysis of the netcode changes in the Spring patch:

    Just wanted to repost this, which I assume is the video Mischkag said everyone should see.
    I didn't watch it when jdb first posted it, will watch it right now.
  • mischkag
    214 postsMember, Developer
    @KingTolapsium: We have not seen or get any new occurrence reported of the hit detection bug. I did fix a few bugs potentially leading up to this, but it is impossible to tell without ever being able to reproduce it :( I'll keep my fingers crossed though.

    @kniphtee: Late joining is in fact a non trivial thing to get absolutely right. Not from an engineering point of view, rather a level design point of view and u need to have networking in your mind. If you could precisely tell me what kind of glitches and bugs occur, i may be able to look into this and push improvements.

    @lizzard: Thank you. Basically if you see the server tick going up may be due to 2 reason: first there is in fact a constant higher load (which would not necessarily be so bad if it stays well below 30ms @ 30Hz) or there are frequent spikes which lead to a higher average. The latter is bad and leads to a lot of fluctuations in the netcode too. Hard to tell and the downside of that is that in order to analyze it, you would have to track a lot of detailed CPU activity which leads to an overall slowdown. The only thing i can do is to monitor this in house and hope to see abnormal activity during our playtests.
    The latency display as well as extrapolation offset really tell you how far you are behind the action (kinda latency/2 + extrapolationOffset). You get the latest server states, the perception of yours may be due to client side frame spikes/low fps perhaps?

  • kniphtee
    244 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member


    [quote="mischkag;c-845643
    @kniphtee: Late joining is in fact a non trivial thing to get absolutely right. Not from an engineering point of view, rather a level design point of view and u need to have networking in your mind. If you could precisely tell me what kind of glitches and bugs occur, i may be able to look into this and push improvements.[/quote]

    The cave/tunnel portion of Monte Grappa (C flag) blanks out as if it is not there (facing the man doors) which makes fire fights from the sides were the doors with a serious problem because you cannot see them - learn to shoot by foot steps I guess.
    Much of the mechanics such as vaulting and jumping seem behind and the animations are lagged as well.
    On Sinai I joined a game midway and spawned into a gas nade on B point so put on my gas mask but the d-pad wasn't working for whatever reason (two week old controller - official xbox stuff) and I died due to gas and the rest of the match every time I spawned back in I had the gas mask on and while the animation showed me taking it off the screen constantly looked like I was still wearing the gas mask. Next map, vote in Scar and all was well as rain. A friend said it's an issue with d-pad input and the game but I haven't been able to replicate - not that I tried hard.
    Lag on team verse enemy icons over the character are common when ads'ing right after spawn into a game but that settles down shortly thereafter.
    Had the black screen glitch recently on dlc maps - mid round join if that matters - later respawn on teammate into nades in hell and the screen went black and spawning all was black - next map right as morning dew.
    The crazy hallucinogenic lights on Edge at G point - only played that map twice in the last week or so but had that happen - there a vids out there on this one.
    When I was finishing the M97 elite codex I had many times in fire fights had the poof of my gun firing on extremely close targets that I assumed would be hit by much of the blast if not inst-kills and it was just pellet poofs without hits which many times resulted in my own death, several times i'd fire off another round or two and end up with trade kills - I guess if I see muzzle blast and what looks like hits to the opponent then I assume in those right conditions for a shotgun i'd have hits - perhaps it is just my poor aim or something else - so be it - maybe random pellet spray to the bullet spray - no se. Next game and timing is right again and I don't have to second glance an enemy I know I dropped.
    On the F flag of Sanai there is a spot on the rocks overlooking F facing towards E there is a black spot amongst the rocks and not like a shadow.
    On Amiens several times I tried to spawn on a team tank (landship) after joining late and each time I would be dropped next to the tank and it wasn't full each of the three times but the one time the spawn dropped me off the bridge on D point to which I died from fall damage - I cried foul, didn't break another controller in my aggravation but it just added to the compounding issue. Next game same teammate in a landship (trying to get his 20 kills or whatever for some medal) on Fao and no problem, jumped right in to his tank and we had a nice streak - even died once ridding our tank of a pesky support trying to limpet us and spawned right back in no problem - well right after that we got bombered but that's not the topic here.
    Can get more footage of these and other things but I just feel bad logging up the xbone with all this video footage and have to mass delete from time to time - let me know thoughts and perhaps some are just perception shaking off the rust but it doesn't feel that way especially after being on a couple times during a play session with each jump back in game just feeling - off as in the timing is not there each time until you start a fresh round at 0.
  • mmarkweII
    2887 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    @kniphtee why don't you record your clips with the netgraph on to show these issues. This thread is about hit detection so you may need to create a new thread or, as suggested in the past, post the link to mischkag twitter so he can see it there. Just my .02.
  • kniphtee
    244 postsMember, Battlefield 3, Battlefield 4, Battlefield, Battlefield 1, CTE Member
    mmarkweII wrote: »
    @kniphtee why don't you record your clips with the netgraph on to show these issues. This thread is about hit detection so you may need to create a new thread or, as suggested in the past, post the link to mischkag twitter so he can see it there. Just my .02.
    Done that multiple times earlier in this thread and others - some still link and some don't - have to clean out the dvr from time to time. Another added note is that these things aren't all hit reg issues but more of game play issues so here i go again going off topic - to me it's all just game issues (off topic again - my bad) but in this case Mischkag inquired. And about the netgraph which i heard there was a smaller one coming but since last patch i have not had a spike over 76 even when the behemoth drops from the sky and i hang consistently around 20-36 - it has become moot for me to have it blocking a quarter of my screen anymore - Spectrum owned up to their 100mbs and have my lines running smooth as silk...and that's a lot of games of cq and ops - credit to @mischkag - mad props (apologies to others who now have probs but i'm running great aside from the jump in game).
  • oJU5T1No
    900 postsMember, Battlefield 3, Battlefield 4, Battlefield Hardline, Battlefield, Battlefield 1 Member
    The de-sync is ridiculous first watch the bullets go straight though this guy, then he throws a grenade after his death.

    http://xboxdvr.com/gamer/oJU5T1No/video/30970531#t=41

  • VBALL_MVP
    4099 postsMember, Battlefield 3, Battlefield 4, Battlefield Hardline, Battlefield, Battlefield 1 Member
    oJU5T1No wrote: »
    The de-sync is ridiculous first watch the bullets go straight though this guy, then he throws a grenade after his death.

    http://xboxdvr.com/gamer/oJU5T1No/video/30970531#t=41

    Looks like you missed. I see the tracers in front of the guy, then when you adjust they are behind. By the time you got on target, he was being killed by the other guy.
Sign In or Register to comment.

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!