Help us reproduce the hit-reg problem

2456

Comments

  • xDragonxDragon Join Date: 2012-04-04 Member: 149948Members, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Shadow
    The twitch animations are not the problem with the hitreg on most alien lifeforms, really only the harvester (and possibly the fade a little bit still).
  • eh?eh? Join Date: 2012-03-03 Member: 147997Members
    I'd say the fade is still like 10-15% miss rate compared to the old 40-60% when shooting at a stationary, 0 ms fade because of flinching.

    The speedhack protection seems like a possible culprit for why skulks / players are warping an unusual amount compared to before it was added, but skulk movement changes could also be to blame... hard to say. And obviously the server / client performance issues as of late aren't helping anything. And beyond that there are definitely some general animation (non-flinch) problems with hitreg, but it's pretty hard to say if it's entirely the engine or the damage representations or just a mix of the two.
  • NeoRussiaNeoRussia Join Date: 2012-08-04 Member: 154743Members
    <!--quoteo(post=2039535:date=Dec 3 2012, 07:56 PM:name=xDragon)--><div class='quotetop'>QUOTE (xDragon @ Dec 3 2012, 07:56 PM) <a href="index.php?act=findpost&pid=2039535"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->The twitch animations are not the problem with the hitreg on most alien lifeforms, really only the harvester (and possibly the fade a little bit still).<!--QuoteEnd--></div><!--QuoteEEnd-->

    Your avatar is what I want to do when this happens (the first clip)

    <a href="http://www.twitch.tv/l2pns2/b/341452008" target="_blank">http://www.twitch.tv/l2pns2/b/341452008</a>

    and it happens a lot.
  • xDragonxDragon Join Date: 2012-04-04 Member: 149948Members, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Shadow
    If you actually watch that video, hes aiming at the corner of the skulk model, and the skulk dies at exact 20 rounds fired, so counting for some spread (the lmg has some, even with the current bugged and small spread), i would say that really isnt all that bad (11 miss, 9 hits). Most of the misses were in the beginning when he was aiming at the upper corner of the model.
  • fanaticfanatic This post has been edited. Join Date: 2003-07-23 Member: 18377Members, Constellation, Squad Five Blue
    <!--quoteo(post=2039445:date=Dec 3 2012, 10:47 PM:name=Max)--><div class='quotetop'>QUOTE (Max @ Dec 3 2012, 10:47 PM) <a href="index.php?act=findpost&pid=2039445"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->It works in our development build (233), but I'll check to see if it is broken in 232.<!--QuoteEnd--></div><!--QuoteEEnd-->
    AFAIK recording isn't broken (it works for me at least), but playback is riddled with bugs and playback speed is entirely reliant on fps, meaning you'll either get fast forward or slow motion playback unless you get identical fps to the computer it was recorded on.
  • stickybootstickyboot Join Date: 2004-01-29 Member: 25711Members, Constellation
    One really easy way to work with youtube videos, is to use the Click2Flash safari plugin. It provides EASY access to the HTML5 video file, usually in mp4 format which is easier to work with. It also saves your laptops battery life. Maybe this will come in handy for not so useful video reports?
    <a href="http://hoyois.github.com/safariextensions/clicktoplugin/" target="_blank">http://hoyois.github.com/safariextensions/clicktoplugin/</a>

    <img src="http://i.imgur.com/0Ypdz.png" border="0" class="linked-image" />

    The other cool thing about this, is that you get frame by frame video navigation in Quicktime (use your arrow keys).

    <img src="http://i.imgur.com/xSiSf.png" border="0" class="linked-image" />
  • douchebagatrondouchebagatron Custom member title Join Date: 2003-12-20 Member: 24581Members, Constellation, Reinforced - Shadow
    i think the problem with youtube is that the video is still compressed, the devs need the raw footage, preferable filmed at a higher than normal FPS. Ideally, if you're running the game at 60 fps, they'd want to review it at 60 fps so they could see the exact frames that were rendered to you. when you upload it to youtube there's a lot of things that change, frames get merged and whatnot.

    the demo records are probably the best thing, provided the devs have a better way to play them back than is available to us. (would really like a way to jump to a spot in the recording, or to slow down, or even cut out parts.)
  • Omar - The WireOmar - The Wire Join Date: 2012-11-01 Member: 165320Members
    <!--quoteo(post=2039558:date=Dec 3 2012, 06:21 PM:name=6john)--><div class='quotetop'>QUOTE (6john @ Dec 3 2012, 06:21 PM) <a href="index.php?act=findpost&pid=2039558"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->the devs need the raw footage<!--QuoteEnd--></div><!--QuoteEEnd-->
    A 45 second video clip, at 30 frames per second, uncompressed is 3.16 GB.

    It also, since it's uncompressed, is almost unplayable.
  • WheeeeWheeee Join Date: 2003-02-18 Member: 13713Members, Reinforced - Shadow
    edited December 2012
    shouldn't the time for this have been 3 weeks ago?
  • nailertnnailertn Join Date: 2012-11-18 Member: 172301Members
    LMG fires 10 bullets per second, youtube videos play at 30 FPS which should be good enough to verify hits and misses. What's the problem with saving the .flv and viewing it in a player that supports frame by frame play?
  • GORGEousGORGEous Join Date: 2012-02-19 Member: 146762Members, NS2 Map Tester
    <!--quoteo(post=2039598:date=Dec 3 2012, 10:04 PM:name=nailertn)--><div class='quotetop'>QUOTE (nailertn @ Dec 3 2012, 10:04 PM) <a href="index.php?act=findpost&pid=2039598"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->LMG fires 10 bullets per second, youtube videos play at 30 FPS which should be good enough to verify hits and misses. What's the problem with saving the .flv and viewing it in a player that supports frame by frame play?<!--QuoteEnd--></div><!--QuoteEEnd-->


    LMG fires ~18 bullets per second.
  • xDragonxDragon Join Date: 2012-04-04 Member: 149948Members, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Shadow
    edited December 2012
    WRONG - DFHJdsok;lk

    Oops wrong number, well not really but posting the true numbers would reveal things that shouldn't be known publicly most likely..

    Effective fire rate when holding down fire continuously is 0.055326889972, so 18 is correct.
  • GORGEousGORGEous Join Date: 2012-02-19 Member: 146762Members, NS2 Map Tester
    <!--quoteo(post=2039645:date=Dec 4 2012, 12:12 AM:name=xDragon)--><div class='quotetop'>QUOTE (xDragon @ Dec 4 2012, 12:12 AM) <a href="index.php?act=findpost&pid=2039645"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->LMG fire rate is actually slightly under 0.08 seconds, so about 12~13 bullets a second, lol.<!--QuoteEnd--></div><!--QuoteEEnd-->

    Where do you get that from?

    Testing in game by unloading an entire 50 bullet mag from LMG I typically get ~2.8 - 3.0s. Nowhere near the 4 seconds 0.08s would yield.
  • nailertnnailertn Join Date: 2012-11-18 Member: 172301Members
    Was looking it up in the wiki, I stand corrected. Nevertheless, 30 FPS still sounds good enough.
  • ComboBreakerComboBreaker Join Date: 2012-11-22 Member: 172856Members
    I wish you best luck with fixing these pesky hitreg issues.Although I have not experienced them yet,it seems that it is a serious problem,judging by uploaded videos.
  • WasabiOneWasabiOne Co-Lead NS2 CDT Join Date: 2011-06-15 Member: 104623Members, NS2 Developer, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Supporter, Reinforced - Silver, Reinforced - Gold, Reinforced - Diamond, Reinforced - Shadow, WC 2013 - Shadow, Subnautica Playtester, Pistachionauts
    FYI in regards to flinch animations, the harvestor indeed "had" a problem and it has already been fixed for 233. The devs are looking into whether or not other entities suffer from the same issue, but it seems that may not be the issue with them. Thank you all for your help in trying to figure this all out! What an awesome community.
  • tk-421tk-421 Join Date: 2006-11-03 Member: 58315Members, Reinforced - Shadow
    <!--quoteo(post=2039581:date=Dec 3 2012, 06:33 PM:name=Wheeee)--><div class='quotetop'>QUOTE (Wheeee @ Dec 3 2012, 06:33 PM) <a href="index.php?act=findpost&pid=2039581"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->shouldn't the time for this have been 3 weeks ago?<!--QuoteEnd--></div><!--QuoteEEnd-->

    Ah, come on. At least it's being addressed.
  • LocklearLocklear [nexzil]kerrigan Join Date: 2012-05-01 Member: 151403Members, NS2 Playtester, NS2 Map Tester, WC 2013 - Shadow
    edited December 2012
    <!--quoteo(post=2039542:date=Dec 3 2012, 04:20 PM:name=xDragon)--><div class='quotetop'>QUOTE (xDragon @ Dec 3 2012, 04:20 PM) <a href="index.php?act=findpost&pid=2039542"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->If you actually watch that video, hes aiming at the corner of the skulk model, and the skulk dies at exact 20 rounds fired, so counting for some spread (the lmg has some, even with the current bugged and small spread), i would say that really isnt all that bad (11 miss, 9 hits). Most of the misses were in the beginning when he was aiming at the upper corner of the model.<!--QuoteEnd--></div><!--QuoteEEnd-->

    Except when I watch it back in Vegas at full resolution in frame by frame I'm seeing blood and I'm not actually missing as far as the client-side model goes.

    This video he linked was a really rough recording that's hard to really show anything.
  • xDragonxDragon Join Date: 2012-04-04 Member: 149948Members, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Shadow
    Rough recording or not you can still see reasonably well.. aiming at the corner of the skulk initially lost you ~5 rounds, if you also notice the firing effects dont start till like 3 or 4 rounds fired so you need to watch the numbers. The skulk dies at 20 shots fired (maybe even 18-19 tbh), so 13-14 bullets to kill IMO, really isnt a bad deal. Id say a couple of those could possibly be lost to hitreg issues, but ive seen much better example of bullets not registering than that.
  • bHackbHack Join Date: 2010-03-23 Member: 71059Members
    edited December 2012
    <a href="http://www.unknownworlds.com/ns2/forums/index.php?s=&showtopic=125040&view=findpost&p=2039705" target="_blank">http://www.unknownworlds.com/ns2/forums/in...t&p=2039705</a>
    <!--quoteo--><div class='quotetop'>QUOTE </div><div class='quotemain'><!--quotec-->However, another high issue is reg, and if someone is telling that it is not 'that bad' - believe me, it is and you are just a bad FPS gamer and have no overall experience at all in NS. The predecessor of NS2 - NS1 had 'such' a good reg (not really until late years:D), but it actually was waaay better than current reg in ns2.

    There are 2 reasons behind it:
    1. NS2 hitboxes stretch/jiggle because of alien skilled 'movement' aka speed boost from wall jump etc, which does not compute the hitbox correctly. Also include ping difference here for threshold. Why would suspect this? Because there are no references to the 'fixes' or any 'adjustments' to the reg in the build changelogs, however the reg feels different compared to 230 and 232, and that's where the major changes to the walljump/speed formulas where applied.

    2. Server tickrate. YES, THAT ONE. Why would the servers with 20-25 tick rate have such BAD BAD reg, whereas 30tickrate servers work 'better'. Because it was and still is the same in HL1/NS/CSS/CS:GO - the higher tickrate is on the server (ABOVE so called 'RECOMMENDED'), the better reg and performance is. NS1 had 30 tickrate servers for long in it's begging stage. With years server hosters improved their hardware etc, and we switched to 60. And than to 100. Even going for 60 drastically improved the reg. And Having 100 tickrate competitive servers compared to 30 tickrate pubs was a good real-time test for reg.
    Back to NS2 - yes, it 'works' with 30 tickrates OK'ish, but not to a satisfactory level.
    Did you ever noticed that client tickrate thing (aka cmdrate)? Yes, it is there (net_stats). 60-70 tickrates? (at least last time I've checked it was around that value. Might go lower in combat 'cos of fps which is expected, but I wasn't paying attention) Well, that's what client side wants to send to the server! However, it can't =( The server will never receive more than its maximum tick rate setting EVER, and basically CAN'T receive OR respond to you with more updates per second than it's tick rate. So why is that important? The higher you set your updaterate, the more accurately your game can show you what's going on (aka hitboxes + reg).

    So what we have in the end with the rates: client is trying to send 60-70+~ tickrates to the server (cmdrate sent, bottlenecked), but the server only computes half of it (or even less), as the server tickrate (in best cases) is 30tickrates (updaterate received).
    Another remark to cmdrate - it is heavily coupled to fps. The higher FPS you have, the higher cmdrate will be generated by your game to send to the server, as it is calculated with each frame rendered.

    Final conclusion - playing on the server that is capable of running reliably at higher tickrates, clients with hardware that can run at higher rates will see MUCH smoother gameplay and hit registration is generally a lot more accurate.
    Generally if the server is running at 60 tickrates, you would set up your updaterate and cmdrate to 60(66). With 100 tickrate servers - to 100(128). Since we can't set this values up, I suspect they are already set to 'good' values (that's why the client tickrate is high at least).
    However, currently the server is bottle-necking with the client.

    Prove me wrong. Go for proper explanations, we can handle them.<!--QuoteEnd--></div><!--QuoteEEnd-->

    Might get some videos after as well.
    Oh, and SG sniping enemies far away with 22/52/72 dmg and doing the same in close range is ridiculousness. And when I say close range I mean my crosshair is just in the middle of a skulk.
    That was never happening in ns1. Skulk would die from 2 close range hits when you point at hear max. I'm not complaining about obvious misses. But obvious straight shots... Especially when the skulks jumps with hear head upfront you, not sidewise.
    close range = 3~5m.
  • Kouji_SanKouji_San Sr. Hινε Uρкεερεг - EUPT Deputy The Netherlands Join Date: 2003-05-13 Member: 16271Members, NS2 Playtester, Squad Five Blue
    <!--quoteo(post=2039903:date=Dec 4 2012, 04:34 PM:name=bHack)--><div class='quotetop'>QUOTE (bHack @ Dec 4 2012, 04:34 PM) <a href="index.php?act=findpost&pid=2039903"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec--><a href="http://www.unknownworlds.com/ns2/forums/index.php?s=&showtopic=125040&view=findpost&p=2039705" target="_blank">http://www.unknownworlds.com/ns2/forums/in...t&p=2039705</a>


    Might get some videos after as well.
    Oh, and SG sniping enemies far away with 22/52/72 dmg and doing the same in close range is ridiculousness. And when I say close range I mean my crosshair is just in the middle of a skulk.
    That was never happening in ns1. Skulk would die from 2 close range hits when you point at hear max. I'm not complaining about obvious misses. But obvious straight shots... Especially when the skulks jumps with hear head upfront you, not sidewise.
    close range = 3~5m.<!--QuoteEnd--></div><!--QuoteEEnd-->
    Like I said the shotgun has very obvious hitreg issues where you need to aim on the edge behind the target (alien) to do a full damage shot with more reliability.

    And there is something funky when an alien is very close and somehow most of your pellets seem to go through the target and not doing damage...
  • NeoRussiaNeoRussia Join Date: 2012-08-04 Member: 154743Members
    <!--quoteo(post=2039906:date=Dec 4 2012, 12:41 PM:name=Kouji_San)--><div class='quotetop'>QUOTE (Kouji_San @ Dec 4 2012, 12:41 PM) <a href="index.php?act=findpost&pid=2039906"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Like I said the shotgun has very obvious hitreg issues where you need to aim on the edge behind the target (alien) to do a full damage shot with more reliability.

    And there is something funky when an alien is very close and somehow most of your pellets seem to go through the target and not doing damage...<!--QuoteEnd--></div><!--QuoteEEnd-->

    This must be why I find the shotgun underpowered as it makes you like a glass cannon from the reload between shots. One-shotting never seems reliable for me even though my shotgun aim is consistent. Long range is a lot easier to hit even if the target is going fast.
  • elmo9000elmo9000 Join Date: 2012-03-24 Member: 149324Members
    <!--quoteo(post=2039906:date=Dec 4 2012, 07:41 PM:name=Kouji_San)--><div class='quotetop'>QUOTE (Kouji_San @ Dec 4 2012, 07:41 PM) <a href="index.php?act=findpost&pid=2039906"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Like I said the shotgun has very obvious hitreg issues where you need to aim on the edge behind the target (alien) to do a full damage shot with more reliability.

    And there is something funky when an alien is very close and somehow most of your pellets seem to go through the target and not doing damage...<!--QuoteEnd--></div><!--QuoteEEnd-->

    Most of the time when close it seems that the aliens go through you, so you end up shooting behind them/inside the hitboxes and it just doesnt reg. Should also finally use some time to fix the collisions/overall physics (lolmarinesclimbingonwalls) as it would help to improve the overall feel of the game a lot.
  • bHackbHack Join Date: 2010-03-23 Member: 71059Members
    edited December 2012
    <!--quoteo(post=2039906:date=Dec 4 2012, 04:41 PM:name=Kouji_San)--><div class='quotetop'>QUOTE (Kouji_San @ Dec 4 2012, 04:41 PM) <a href="index.php?act=findpost&pid=2039906"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Like I said the shotgun has very obvious hitreg issues where you need to aim on the edge behind the target (alien) to do a full damage shot with more reliability.

    And there is something funky when an alien is very close and somehow most of your pellets seem to go through the target and not doing damage...<!--QuoteEnd--></div><!--QuoteEEnd-->

    << Yesterday didn't kill cara-skulk with 4 shots :D 2 of them were straight in the middle of its model.
  • Kouji_SanKouji_San Sr. Hινε Uρкεερεг - EUPT Deputy The Netherlands Join Date: 2003-05-13 Member: 16271Members, NS2 Playtester, Squad Five Blue
    edited December 2012
    <!--quoteo(post=2039908:date=Dec 4 2012, 04:49 PM:name=elmo9000)--><div class='quotetop'>QUOTE (elmo9000 @ Dec 4 2012, 04:49 PM) <a href="index.php?act=findpost&pid=2039908"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Most of the time when close it seems that the aliens go through you, so you end up shooting behind them/inside the hitboxes and it just doesnt reg. Should also finally use some time to fix the collisions/overall physics (lolmarinesclimbingonwalls) as it would help to improve the overall feel of the game a lot.<!--QuoteEnd--></div><!--QuoteEEnd-->
    Yep, its like the aliens go behind your shotgun pellet origin or perhaps the spread pattern up close is bigger then the target on screen in some inverted kind of way. Hell I dunno, it's odd in any case... I do get consistent one shot kills though, but you have to force yourself to ignore the model position and shoot at the spot behind the target :/

    I find the shotty to work best in a range in between 1.5-2m and ~3-4m



    Also I hate seeing marines on my walls dammit, that is my domain :P
  • PimpToadPimpToad Join Date: 2012-11-02 Member: 166005Members
    No video, but I can share an experience I found rather amusing. I've recently taken to using the Shotgun as I can't seem to hit Skulks these days with my AR. Imagine my surprise when I'm constantly getting 20 damage (sometimes 40) at rather close range. I'd say this is around <3m mark. The most amusing example came in Tram on the Repair power node if I recall correctly. A skulk was chewing on the power node completely oblivious to my presence. I moved to around <3m range and centered my crosshair on its body. A burst of green goo later and I'm rewarded with 20 damage according to the display. Fast forward a minute or so later and another skulk was yet again chewing on the power node oblivious to my presence. I moved to almost the same exact position at <3m range and fired the same shot. Lo and behold it instantly gibbed the poor thing with around 180something damage being displayed. I dunno if this is more amusing than Onos goring my face while my Shotgun is doing 20/40 damage at point blank range while I get 180+ when the space cow is half a room away at full retreat.

    I would love to record a video at 60fps to show some of these shenanigans, but can't be helped when my fps just love to dip to the low 20s when in combat. Though I take comfort in people with better rigs than I do having the same issue.
  • rantologyrantology Join Date: 2012-02-05 Member: 143750Members, NS2 Developer, NS2 Playtester, Squad Five Gold, NS2 Map Tester, Reinforced - Shadow, WC 2013 - Gold
    Shotgun is probably not the best weapon to use to document hitreg issues, the <a href="http://i.imgur.com/Uv3dU.jpg" target="_blank">spread</a> is incredibly RNG.
  • ZekZek Join Date: 2002-11-10 Member: 7962Members, NS1 Playtester, Constellation, Reinforced - Shadow
    edited December 2012
    I don't think they're saying that Youtube video quality is insufficient, but rather that it doesn't have the tools to examine them frame by frame. If that's the case, I recommend just downloading the FLVs from Youtube(there are lots of apps to do that) and finding a better FLV player to view them in.
  • Kouji_SanKouji_San Sr. Hινε Uρкεερεг - EUPT Deputy The Netherlands Join Date: 2003-05-13 Member: 16271Members, NS2 Playtester, Squad Five Blue
    edited December 2012
    The demo files have a lot of engine related tools attached to them, to analyse the code running at the time. That is why they prefer the in engine demos. The demo recorder was initially fixed up as a tool to analyse bugs with possibilities extending that into a proper demo recorder for better replays.
  • bERt0rbERt0r Join Date: 2005-03-23 Member: 46181Members
    I just had an idea: what if the prediction makes skulks look like they jump higher than they actually do?
Sign In or Register to comment.