<!--quoteo(post=2024622:date=Nov 15 2012, 07:40 AM:name=BloodyIron)--><div class='quotetop'>QUOTE (BloodyIron @ Nov 15 2012, 07:40 AM) <a href="index.php?act=findpost&pid=2024622"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Back to NS2, the issue is hit detection with the machine gun. As my original example, the marine turned around and in less than half a second tore me down to exactly 1 hitpoint, when I was at full health and armor. Would anyone actually declare this fair, balanced and working as intended? I sure dont.<!--QuoteEnd--></div><!--QuoteEEnd-->
I don't get it either. On Summit, sitting at max range on any of the hundreds of random ledges a lerk can sit on in crevice, I've shot maybe 4 needles worth of spikes (silenced spikes no less) at a marine coming out of the Flight Control door, and have been killed before visually even being able to see the marine in question turning towards me. And it's not just a fluke either, has happened several times with all the parameters changed (server, opponent, ledge, angle) and it really grinds my gears that I can be smoked by someone who visually has their back towards me.
There is definitely a problem in the game. I hope they fix it. Plenty of times as an alien I'll be getting shot, so I hop around the corner and I'm already a good 5 yards down the hall around the corner and the marine which is not even remotely within line of sight at this point still kills me "through the wall". I know a LOT of people have been complaining about it.
<!--quoteo(post=2024633:date=Nov 15 2012, 12:52 PM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 12:52 PM) <a href="index.php?act=findpost&pid=2024633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Im sorry, but is there really such thing as competitive TF2??? what the hell?<!--QuoteEnd--></div><!--QuoteEEnd-->
Youtube is your friend. and now I'll have to watch Axefane again =|
<!--quoteo(post=2024634:date=Nov 15 2012, 07:53 AM:name=Metal Handkerchief)--><div class='quotetop'>QUOTE (Metal Handkerchief @ Nov 15 2012, 07:53 AM) <a href="index.php?act=findpost&pid=2024634"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I don't get it either. On Summit, sitting at max range on any of the hundreds of random ledges a lerk can sit on in crevice, I've shot maybe 4 needles worth of spikes (silenced spikes no less) at a marine coming out of the Flight Control door, and have been killed before visually even being able to see the marine in question turning towards me. And it's not just a fluke either, has happened several times with all the parameters changed (server, opponent, ledge, angle) and it really grinds my gears that I can be smoked by someone who visually has their back towards me.<!--QuoteEnd--></div><!--QuoteEEnd--> there's something weird with the way it updates - I've had a rifle bring me from 70 to 20 hp with no updates in between. not really sure what the cause of this one is, could just be network lag or something
<!--quoteo(post=2024633:date=Nov 15 2012, 07:52 AM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 07:52 AM) <a href="index.php?act=findpost&pid=2024633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Im sorry, but is there really such thing as competitive TF2??? what the hell?<!--QuoteEnd--></div><!--QuoteEEnd--> more competitive than competitive NS2.
IronHorseDeveloper, QA Manager, Technical Support & contributorJoin Date: 2010-05-08Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
That turning speed is not hit reg but smoothing. start a server, go third person and observe your twitch aiming slightly delayed. You can change this value to test different levels of smoothing, but remember that u need some..
Also for what its worth increasing the tick rate would only effect ai like macs, according to Max.. And lowering the interp would require increasing network traffic, potentially throwing server and client performance backwards. I.e. optimize first then improve these settings.
I noticed that your fire rate is also substantially slower as well as hit registry with lower framerate. Not really okay for a competitive shooter, unless they improve optimization real fast.
Team Fortress 2 has had tournaments and competitions for it since the year it came out. It has online tournaments and LAN tournaments, as well as other obscure tournaments throughout the world. In fact there was a recent tournament where the NA TF2 scene sent one of their teams to EU to do a battle of the continents.
I personally have entered tournaments and leagues run by the following organizations: -TWL -CAL/CPL -ESEA -CEVO -Lanageddon -Fragapalooza
I also run an organization that has developed a competitive format that has been picked up by _other_ organizations and had tournaments using the format, my group is LANified!, and the group that provides the largest coverage for TF2 content picked up our format: <a href="http://www.extelevision.com/saltv-3v3-koth-tournament/" target="_blank">http://www.extelevision.com/saltv-3v3-koth-tournament/</a>
There also have been times where TF2 was the most played game on STEAM, and later times when it kept in the top 3. Currently it stays steadily in top 10, and they keep releasing original content for it.
<!--quoteo(post=2024633:date=Nov 15 2012, 09:52 AM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 09:52 AM) <a href="index.php?act=findpost&pid=2024633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Im sorry, but is there really such thing as competitive TF2??? what the hell?<!--QuoteEnd--></div><!--QuoteEEnd-->
Whatever the source(s) of the problem, it is my <i>perception </i>that there are indeed times when I die after getting behind cover, or get burst from 100% to death with no intermediate state. Now, even if it's just perception, you still have to address that, or else players global experience is diminished. You have to address it differently then if there was a "real" problem, but it still needs to be addressed. Perception is one of the more important aspects of your global enjoyment of games. Wathever magic works (or doesn't) behind the code-curtain, all you know is what you perceive.
And I definitely feel that sometimes, the game doesn't play "fairly", in the sense that I am getting hit after being in an "untouchable" situation (behind cover or gone).
With that said, it is understandable that no game can be perfect when it comes to lag/timing/syncronisation and perception. But, I do feel that there is room for improvement. I hope so anyways!
*Btw, when this happens, I'm playing with a sub 75 ping.
<!--quoteo(post=2024879:date=Nov 15 2012, 03:41 PM:name=ironhorse)--><div class='quotetop'>QUOTE (ironhorse @ Nov 15 2012, 03:41 PM) <a href="index.php?act=findpost&pid=2024879"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Need video of that neorussia<!--QuoteEnd--></div><!--QuoteEEnd-->
I can't actually, because my fps would be like 10 when recording. I could record in a local server, but the fact that my fps is lower in the first place would kinda alter the results.
Yea it's funny that someone asked if TF2 was competitive, because it's still the only teamwork-based shooter in competition, and it having seasons of high money value LANs and a very high skill ceiling for several years now with over a hundred NA teams of more than 6 in ESEA and even a higher amount of European teams. NS also has leaked most of its best players to TF2 like wltrs and some players that were on team xensity to name a few. In the eyes of these people it is highly unlikely that NS2 will ever come close to either of these games because of the quality of design decisions.
NS2's netcode is really carrior pigeons that transfer your packets back and forth...
It all seriousness however, for a competitive game the interpolation in NS2 is too high. But IMO that is the least of the problems currently, the poor movement, collisions, hit detection and low FPS all hold this game back immensly, and while the devs are and have been working on fixing them, many issues still remain. IMO hit detection is extremely broken currently, there is some major issues there which needs fixing. IMO a really good way to show the problem is to shoot a gliding lerk as it flies by quickly... You can start landing hits correctly and then randomly all your shots start missing, even when your tracking it perfectly. You can test this on a lerk flying in a curved line and its usually pretty reproduceable.
<!--quoteo(post=2024886:date=Nov 15 2012, 11:50 AM:name=Uh-Oh)--><div class='quotetop'>QUOTE (Uh-Oh @ Nov 15 2012, 11:50 AM) <a href="index.php?act=findpost&pid=2024886"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Whatever the source(s) of the problem, it is my <i>perception </i>that there are indeed times when I die after getting behind cover, or get burst from 100% to death with no intermediate state. Now, even if it's just perception, you still have to address that, or else players global experience is diminished. You have to address it differently then if there was a "real" problem, but it still needs to be addressed. Perception is one of the more important aspects of your global enjoyment of games. Wathever magic works (or doesn't) behind the code-curtain, all you know is what you perceive.
And I definitely feel that sometimes, the game doesn't play "fairly", in the sense that I am getting hit after being in an "untouchable" situation (behind cover or gone).
With that said, it is understandable that no game can be perfect when it comes to lag/timing/syncronisation and perception. But, I do feel that there is room for improvement. I hope so anyways!
*Btw, when this happens, I'm playing with a sub 75 ping.<!--QuoteEnd--></div><!--QuoteEEnd--> another thing to remember on top of all this netcode discussion is where the camera is in the Skulk's model
this can lead to you thinking you're in cover when your front claws are peeking out a vent, or being shot around a corner when your back feet haven't passed the corner yet
Interesting that Quake 3 was mentioned. Q3 was one of the last "legacy" games that did hit detection naively and server-side. When a "fire" command came in, it did the hit detection against its (the server's) perception of the game world, and that was what counted. This perception was, of course, different from either player, since the firing player thought the target was a behind where the server thought, and the target themselves thought they were ahead of where the server thought.
There was a mod to correct this, Unlagged (http://www.ra.is/unlagged/). It modernized the Quake 3 netcode in many ways, and I believe it predated the implementation of latency correction in most other games. It did what has been described in previous posts in this thread; it rolled back player positions based on the firing player's latency, and THEN did the hit detection.
I was involved a bit in the testing of the mod, and I remember having discussions with the author about latency correction on projectile weapons. Hitscan weapons (instant-hit basically) is easy, but what about when you fire a very rapidly moving projectile like a rocket? The author stuck latency correction for this into a later build, and we tried it out, but there was a notable side effect; the rocket's path was dependent on the starting position of the player firing it, so the normal latency correction approach didn't work; every other player would see the rocket moving on a different path/direction than the firing player and server would.
The end solution was to start the rocket, in everybody else's perspective, from where it was when the other players "found out" about the rocket. In other words, if I fired a rocket, you would see it appear multiple feet in FRONT of me, rather than appearing at the mouth of my rocket launcher. It was not a great solution, but it was the only one that seemed to work.
I'd actually be interested to know how modern game engines like Spark handle the fast-moving-non-hitscan-projectile problem.
EDIT: It looks like later versions of unlagged went with a hybrid approach where he basically bypassed the 50ms of interpolation in certain cases for missiles, letting him get away with up to 50ms of "side-effect free" latency correction on missiles.
<!--quoteo(post=2024922:date=Nov 15 2012, 04:32 PM:name=Guspaz)--><div class='quotetop'>QUOTE (Guspaz @ Nov 15 2012, 04:32 PM) <a href="index.php?act=findpost&pid=2024922"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I'd actually be interested to know how modern game engines like Spark handle the fast-moving-non-hitscan-projectile problem.<!--QuoteEnd--></div><!--QuoteEEnd-->
well, most popular games have engines derivative of idtech engines. CoD and all source games for example. In Spark, the grenade launcher is an atrocity. It's impossible to aim with it because of the horrible interpolation and the way grenades like to fly into your face because that marine that moved through your view 10 seconds ago counts as being in your way because that's what the server thinks. Thanks UWE I always wanted to spend resources just so I can kill myself and lose my jetpack.
My favourite part of the engine is how an onos with 100 ping can hit you all the way across lava falls (literally, it has happened) and a skulk of similar ping would always be able to bite you before he is in one-shot range of your shotgun. That's some QUALITY netcode.
Ill be checking out youtube then. I just never thought a game like TF2 could make it to a competitive field.
it is one of those "turn your brain of and click" games. IMO the first one had more depth than this one (or it feels taht way because it was slower, or because I was younger?), but what the hell, if it is being played competitively I may have to check it out.
AAAAnd on topic.
Neo, Ive had that happen with the grenade launcher as well, I thoguht I had hit a wall or something, but that makes more sense! The grenade launcher is too unpredictable, it makes it not fun to use.
<!--quoteo(post=2024970:date=Nov 15 2012, 04:59 PM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 04:59 PM) <a href="index.php?act=findpost&pid=2024970"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Ill be checking out youtube then. I just never thought a game like TF2 could make it to a competitive field.
it is one of those "turn your brain of and click" games. IMO the first one had more depth than this one (or it feels taht way because it was slower, or because I was younger?)<!--QuoteEnd--></div><!--QuoteEEnd-->
TFC and QTF are completely different from TF2 in competitive play. My favourite movie is Fanom <a href="http://www.youtube.com/watch?v=ENruOrM01CQ&hd=1" target="_blank">http://www.youtube.com/watch?v=ENruOrM01CQ&hd=1</a> as it shows actual teamplay. Although they aren't the best, I do have videos of my own. They are found on my youtube channel in my signature.
- Marine sees you. - 50ms - his 50ms internet latency - Server receives packet of marine seeing you. - 82ms - 32ms based on tick rate - Server is done processing marine seeing you. - 132ms - your 50ms internet latency - Your computer receives packet of marine seeing you. (meanwhile marine starts shooting you around here if he wasn't already pre-firing) - 232ms - 100ms interpolation - Your computer is done waiting for data to interpolate between. - 292ms - 60ms render time - Your computer can easily take this long to render a frame depending on your settings and frame rate. - 307ms - 15ms monitor latency - Monitors often have some processing lag. It's smaller on TN displays, but this is common. You now see marine. - 507ms - 200ms human reaction time - 515ms - 8ms default mouse frequency - 531ms - guessing on a 16ms processing time here; I don't think it waits for the frame before sending data to server? - 581ms - your 50ms latency - Server receives your movement to get around the wall - 631ms - guessing on 50ms to move out of the way - 663ms - 32ms server processing time - 713ms - his 50ms internet latency - 813ms - 100ms interpolation - 829ms - guessing again at 16ms processing time (here I THINK the marine can't hit you anymore even though he still hasn't caught up visually)
I'm just demonstrating how things add up. That's like 629ms of him shooting you. I'm probably wrong on some of that (hard to get the order right, maybe someone can do better), but my point is made I hope. Because skulks move so fast, it accentuates all these latencies. You can do SOME things to reduce the problem, though. Turn off vsync,get a higher FPS, turn off/down frame pre-rendering, stay on servers with low latencies, get a low latency monitor, make sure you're using a high frequency mouse. As far as Unknown Worlds, they can improve client and server performance, reduce the interpolation period (67ms would be better, I think, for this tick rate), up the tick rate once server performance improves.
A lot of that actually happens in parallel, I'm pretty sure your calculations are way off too, almost 1s delay is definitely not what is happening.
<!--quoteo(post=2025088:date=Nov 15 2012, 05:12 PM:name=PseudoKnight)--><div class='quotetop'>QUOTE (PseudoKnight @ Nov 15 2012, 05:12 PM) <a href="index.php?act=findpost&pid=2025088"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->(approximate best guesses based on what I know)
- Marine sees you. - 50ms - his 50ms internet latency - Server receives packet of marine seeing you. - 82ms - 32ms based on tick rate - Server is done processing marine seeing you. - 132ms - your 50ms internet latency - Your computer receives packet of marine seeing you. (meanwhile marine starts shooting you around here if he wasn't already pre-firing) - 232ms - 100ms interpolation - Your computer is done waiting for data to interpolate between. - 292ms - 60ms render time - Your computer can easily take this long to render a frame depending on your settings and frame rate. - 307ms - 15ms monitor latency - Monitors often have some processing lag. It's smaller on TN displays, but this is common. You now see marine. - 507ms - 200ms human reaction time - 515ms - 8ms default mouse frequency - 531ms - guessing on a 16ms processing time here; I don't think it waits for the frame before sending data to server? - 581ms - your 50ms latency - Server receives your movement to get around the wall - 631ms - guessing on 50ms to move out of the way - 663ms - 32ms server processing time - 713ms - his 50ms internet latency - 813ms - 100ms interpolation - 829ms - guessing again at 16ms processing time (here I THINK the marine can't hit you anymore even though he still hasn't caught up visually)
I'm just demonstrating how things add up. That's like 629ms of him shooting you. I'm probably wrong on some of that (hard to get the order right, maybe someone can do better), but my point is made I hope. Because skulks move so fast, it accentuates all these latencies. You can do SOME things to reduce the problem, though. Turn off vsync,get a higher FPS, turn off/down frame pre-rendering, stay on servers with low latencies, get a low latency monitor, make sure you're using a high frequency mouse. As far as Unknown Worlds, they can improve client and server performance, reduce the interpolation period (67ms would be better, I think, for this tick rate), up the tick rate once server performance improves.<!--QuoteEnd--></div><!--QuoteEEnd-->
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=2025991:date=Nov 16 2012, 07:46 PM:name=BloodyIron)--><div class='quotetop'>QUOTE (BloodyIron @ Nov 16 2012, 07:46 PM) <a href="index.php?act=findpost&pid=2025991"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->A lot of that actually happens in parallel, I'm pretty sure your calculations are way off too, almost 1s delay is definitely not what is happening.<!--QuoteEnd--></div><!--QuoteEEnd-->
Mmm... its pretty accurate, actually. There are some diffs, but they are basically a wash.
Its the same in pretty much all lag-compensated games, its just that its not as noticeable in slower-paced, longer ranged games. In NS2, 800 ms allows a celerity/wall-jumping skulk to move 10m, so you can be dead before you start moving. In CoD, 800ms is less than a third of that.
The reason you don't experience it as an 1s delay is that the 1s is that its how long time it takes for a targets reaction to propagate to the attackers world. Get a surprise attack on a marine and try to measure how long time it takes for that marine to start dodging your attack; it will be pretty close to 0.75-1sec or so.
As a marine only needs .4 second of rifle fire to kill a skulk, any skulk waiting until he gets hit before starting to move is already dead.
This effect is most noticeable when you move back around corners and I see it complained about even in slower games like CS. Good players in Counter-Strike, peak corners on other players first, rather than wait for the other player to peak the corner. This gives them a few hundred millisecond advantage. But since skulks are all melee, there is less advantage to peaking them because you can't kill them from range. However, if you leap around a corner, this will give you an extra bit of time to close the distance. Another way, if you're trying to spot or tag them, is to repeatedly peak out for as short a time as possible. This will reduce the time they have to shoot you compared to just sitting there. It's still possible for them to kill you in that time, but they'd have to be ready for you. That's why you might not want to peak the second time if you miss the first time.
But this is all if you're not doing what you should be doing if you can help it -- get behind them.
<!--quoteo(post=0:date=:name=BearTornado)--><div class='quotetop'>QUOTE (BearTornado)</div><div class='quotemain'><!--quotec-->There's a 100ms delay or something along those lines that can screw over aliens who aren't moving when a marine first shoots at them ...<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->The delay is uncertain but the statement is true, and totally unacceptable.<!--colorc--></span><!--/colorc-->
<!--quoteo(post=0:date=:name=Metal Handkerchief)--><div class='quotetop'>QUOTE (Metal Handkerchief)</div><div class='quotemain'><!--quotec-->... and it really grinds my gears that I can be smoked by someone who visually has their back towards me ...<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->Plays off the quote above, and to me, is unacceptable. My play style relies a lot on predicting player moment and patterns, as it should be for just about any player. With this 'delay' or misrepresented model update, movement prediction is almost entirely thrown out the window. And this isn't to mention the teleporting player models ...<!--colorc--></span><!--/colorc-->
<!--quoteo(post=0:date=:name=Swiftspear)--><div class='quotetop'>QUOTE (Swiftspear)</div><div class='quotemain'><!--quotec-->... For the most part we've had backwards progress in gaming in the last several years.<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->I couldn't have said it better myself ... well, maybe a little emphasis "<i><b>A LOT</b> of backwards progress</i>"<!--colorc--></span><!--/colorc-->
<!--quoteo(post=0:date=:name=xDragon)--><div class='quotetop'>QUOTE (xDragon)</div><div class='quotemain'><!--quotec-->... IMO that is the least of the problems currently, the poor movement, collisions, hit detection and low FPS all hold this game back immensly, and while the devs are and have been working on fixing them, many issues still remain. IMO hit detection is extremely broken currently, there is some major issues there which needs fixing. IMO a really good way to show the problem is to shoot a gliding lerk as it flies by quickly... You can start landing hits correctly and then randomly all your shots start missing, even when your tracking it perfectly. You can test this on a lerk flying in a curved line and its usually pretty reproduceable ...<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->Agreed 100%
I've been playing FPS for about 17 years now, and other genres for much longer. <a href="https://developer.valvesoftware.com/wiki/Lag_compensation" target="_blank"><i>Lag Compensation</i></a> had it's place back in late 1990's and early 2000's when most of the playing field was 33.6 - 56kbit/s (general population had a 150+ ping) and broadband was just making its debut. In regards to the idea itself, it was a great idea, but just like most things these days, it was good in writing and not in practice.
In my opinion, Lag Compensation should be taken out entirely. I feel it has <i>NEVER</i> worked correctly and give the games that use it a pseudo feel to it. This feeling generally stems from my experience of pre-Lag Compensation days or what I like to call true net code, where Lag Compensation was done by you and hit detection was greater than 99%.
According to <a href="http://www.verizonbusiness.com/about/network/latency/" target="_blank">Verizon</a>, average latency times across the US are 45ms or less. In my experience, if you play with that ping, you can use insta-hit weapons without compensating in almost all situations.
In summary, Lag Compensation, IMO, is no longer needed and adds unnecessary complexity to net code; which still (in 14 years now) doesn't work correctly. In combination with misrepresenting model position and prediction timing on the player's part, is an absolute fail solution.<!--colorc--></span><!--/colorc-->
<!--quoteo(post=2027214:date=Nov 17 2012, 03:22 PM:name=NSDigi)--><div class='quotetop'>QUOTE (NSDigi @ Nov 17 2012, 03:22 PM) <a href="index.php?act=findpost&pid=2027214"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->... Lag Compensation should be taken out entirely ...<!--colorc--></span><!--/colorc--><!--QuoteEnd--></div><!--QuoteEEnd-->
IronHorseDeveloper, QA Manager, Technical Support & contributorJoin Date: 2010-05-08Member: 71669Members, Super Administrators, Forum Admins, Forum Moderators, NS2 Developer, NS2 Playtester, Squad Five Blue, Subnautica Playtester, Subnautica PT Lead, Pistachionauts
<!--quoteo(post=2024895:date=Nov 15 2012, 12:02 PM:name=NeoRussia)--><div class='quotetop'>QUOTE (NeoRussia @ Nov 15 2012, 12:02 PM) <a href="index.php?act=findpost&pid=2024895"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I can't actually, because my fps would be like 10 when recording. I could record in a local server, but the fact that my fps is lower in the first place would kinda alter the results.<!--QuoteEnd--></div><!--QuoteEEnd-->
Use the record command in the console it will have almost zero impact on performance. Then when you play it back you can use a screen capture program to illustrate the issue you speak of when typically playing.
This would be most helpful for you to accomplish, and try to have r_stats on when you do this. Also i think a key (spacebar?) plays the demo back at actual fps recorded instead of as fast as your machine can render the demo.
ScardyBobScardyBobJoin Date: 2009-11-25Member: 69528Forum Admins, Forum Moderators, NS2 Playtester, Squad Five Blue, Reinforced - Shadow, WC 2013 - Shadow
<!--quoteo(post=2027214:date=Nov 17 2012, 01:22 PM:name=NSDigi)--><div class='quotetop'>QUOTE (NSDigi @ Nov 17 2012, 01:22 PM) <a href="index.php?act=findpost&pid=2027214"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->According to <a href="http://www.verizonbusiness.com/about/network/latency/" target="_blank">Verizon</a>, average latency times across the US are 45ms or less. In my experience, if you play with that ping, you can use insta-hit weapons without compensating in almost all situations.<!--QuoteEnd--></div><!--QuoteEEnd--> 0_o, I'm calling BS on this one. Unless you have a lucky connection to a major hub, you're going to get something like 20-30ms just getting out of your residential connection.
Here's my connection to a server just over 10 miles away <a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459555.png" border="0" class="linked-image" /></a>
And two across the country for me (I'm in Seattle) <a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459519.png" border="0" class="linked-image" /></a> <a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459739.png" border="0" class="linked-image" /></a>
Unless I play only on servers within the Pacific Northwest, there is no way I'm averaging on 45ms.
Also, it looks like the Verizon data is for enterprise (think server farms) connections, not residential. I'd double or even triple those latency values to account for residential connections.
I don't think you understand what I'm saying. The record feature is pretty much broken like a lot of other things for the same reason such as hit registry and key input, because it's based on client-side performance and framerate which at this state of the game is horribly optimized. So even if you could run a demo at the same speed as a game on a server it would still be very stuttery (I've tried space and everything else, changed nothing). I have already tried giving a demo of my own game to a friend with a beast rig and he had the same problems using a demo recorded by me. If the game was optimized to run well it would still alter the results of testing firing speed though. I hope a proper replay system comes soon and all of these stupid engine bugs/limitations get squashed. Just for kicks I'm going to upload one anyway.
<!--quoteo(post=2026017:date=Nov 16 2012, 02:19 PM:name=matso)--><div class='quotetop'>QUOTE (matso @ Nov 16 2012, 02:19 PM) <a href="index.php?act=findpost&pid=2026017"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Mmm... its pretty accurate, actually. There are some diffs, but they are basically a wash.
Its the same in pretty much all lag-compensated games, its just that its not as noticeable in slower-paced, longer ranged games. In NS2, 800 ms allows a celerity/wall-jumping skulk to move 10m, so you can be dead before you start moving. In CoD, 800ms is less than a third of that.
The reason you don't experience it as an 1s delay is that the 1s is that its how long time it takes for a targets reaction to propagate to the attackers world. Get a surprise attack on a marine and try to measure how long time it takes for that marine to start dodging your attack; it will be pretty close to 0.75-1sec or so.
As a marine only needs .4 second of rifle fire to kill a skulk, any skulk waiting until he gets hit before starting to move is already dead.
Which is why good players never stops moving.<!--QuoteEnd--></div><!--QuoteEEnd-->
Competitive games require higher tick rates, i believe it's 30 at the moment for NS2? CS:GO has 64 on their competitive match making servers and it's pretty bad to say the least, keep in mind CS:GO is a slower game - in terms of movement speed - than NS2, so you can only imagine how bad 30 ticks is for a fast game like NS2. But I believe 64 ticks would be enough for NS2 since you don't really need to aim at small hit boxes - headshot -, so i'm assuming as performance improves, they'll increase the tick rate, right?
<!--quoteo(post=2027463:date=Nov 17 2012, 06:16 PM:name=NeoRussia)--><div class='quotetop'>QUOTE (NeoRussia @ Nov 17 2012, 06:16 PM) <a href="index.php?act=findpost&pid=2027463"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->I don't think you understand what I'm saying. The record feature is pretty much broken like a lot of other things for the same reason such as hit registry and key input, because it's based on client-side performance and framerate which at this state of the game is horribly optimized. So even if you could run a demo at the same speed as a game on a server it would still be very stuttery (I've tried space and everything else, changed nothing). I have already tried giving a demo of my own game to a friend with a beast rig and he had the same problems using a demo recorded by me. If the game was optimized to run well it would still alter the results of testing firing speed though. I hope a proper replay system comes soon and all of these stupid engine bugs/limitations get squashed. Just for kicks I'm going to upload one anyway.<!--QuoteEnd--></div><!--QuoteEEnd--> thisssssssssssssssssssssss
I know that UWE has addressed the replay system lacking features, but have they addressed the fact that it literally doesn't work at all?
1) you can't view a demo if it was recorded in a different resolution 2) if I put my graphics settings to high it moves in complete slow motion because it just doesn't run 3) if I try to watch it on the *same* settings that I recorded it, it still skips and lags
<!--quoteo(post=2027214:date=Nov 17 2012, 02:22 PM:name=NSDigi)--><div class='quotetop'>QUOTE (NSDigi @ Nov 17 2012, 02:22 PM) <a href="index.php?act=findpost&pid=2027214"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->In my opinion, Lag Compensation should be taken out entirely. I feel it has <i>NEVER</i> worked correctly and give the games that use it a pseudo feel to it. This feeling generally stems from my experience of pre-Lag Compensation days or what I like to call true net code, where Lag Compensation was done by you and hit detection was greater than 99%.<!--QuoteEnd--></div><!--QuoteEEnd-->So your solution is to replace a subtle visual discrepancy that only a few people notice with a gigantic visual discrepancy that everyone notices? Lag compensation levels the playing field and improves the experience for most players. Improvements can be made, but they usually go hand in hand with performance optimizations, which have been slow going with the lua layer overhead. (as far as I understand it) NS1, iirc, had even worse server performance at release. Let's not freak out here.
Comments
I don't get it either. On Summit, sitting at max range on any of the hundreds of random ledges a lerk can sit on in crevice, I've shot maybe 4 needles worth of spikes (silenced spikes no less) at a marine coming out of the Flight Control door, and have been killed before visually even being able to see the marine in question turning towards me. And it's not just a fluke either, has happened several times with all the parameters changed (server, opponent, ledge, angle) and it really grinds my gears that I can be smoked by someone who visually has their back towards me.
Youtube is your friend.
and now I'll have to watch Axefane again =|
there's something weird with the way it updates - I've had a rifle bring me from 70 to 20 hp with no updates in between. not really sure what the cause of this one is, could just be network lag or something
<!--quoteo(post=2024633:date=Nov 15 2012, 07:52 AM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 07:52 AM) <a href="index.php?act=findpost&pid=2024633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Im sorry, but is there really such thing as competitive TF2??? what the hell?<!--QuoteEnd--></div><!--QuoteEEnd-->
more competitive than competitive NS2.
Also for what its worth increasing the tick rate would only effect ai like macs, according to Max..
And lowering the interp would require increasing network traffic, potentially throwing server and client performance backwards. I.e. optimize first then improve these settings.
I personally have entered tournaments and leagues run by the following organizations:
-TWL
-CAL/CPL
-ESEA
-CEVO
-Lanageddon
-Fragapalooza
I also run an organization that has developed a competitive format that has been picked up by _other_ organizations and had tournaments using the format, my group is LANified!, and the group that provides the largest coverage for TF2 content picked up our format: <a href="http://www.extelevision.com/saltv-3v3-koth-tournament/" target="_blank">http://www.extelevision.com/saltv-3v3-koth-tournament/</a>
There also have been times where TF2 was the most played game on STEAM, and later times when it kept in the top 3. Currently it stays steadily in top 10, and they keep releasing original content for it.
<!--quoteo(post=2024633:date=Nov 15 2012, 09:52 AM:name=DamDSx)--><div class='quotetop'>QUOTE (DamDSx @ Nov 15 2012, 09:52 AM) <a href="index.php?act=findpost&pid=2024633"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->Im sorry, but is there really such thing as competitive TF2??? what the hell?<!--QuoteEnd--></div><!--QuoteEEnd-->
Now, even if it's just perception, you still have to address that, or else players global experience is diminished. You have to address it differently then if there was a "real" problem, but it still needs to be addressed. Perception is one of the more important aspects of your global enjoyment of games.
Wathever magic works (or doesn't) behind the code-curtain, all you know is what you perceive.
And I definitely feel that sometimes, the game doesn't play "fairly", in the sense that I am getting hit after being in an "untouchable" situation (behind cover or gone).
With that said, it is understandable that no game can be perfect when it comes to lag/timing/syncronisation and perception. But, I do feel that there is room for improvement. I hope so anyways!
*Btw, when this happens, I'm playing with a sub 75 ping.
I can't actually, because my fps would be like 10 when recording. I could record in a local server, but the fact that my fps is lower in the first place would kinda alter the results.
Yea it's funny that someone asked if TF2 was competitive, because it's still the only teamwork-based shooter in competition, and it having seasons of high money value LANs and a very high skill ceiling for several years now with over a hundred NA teams of more than 6 in ESEA and even a higher amount of European teams. NS also has leaked most of its best players to TF2 like wltrs and some players that were on team xensity to name a few. In the eyes of these people it is highly unlikely that NS2 will ever come close to either of these games because of the quality of design decisions.
It all seriousness however, for a competitive game the interpolation in NS2 is too high. But IMO that is the least of the problems currently, the poor movement, collisions, hit detection and low FPS all hold this game back immensly, and while the devs are and have been working on fixing them, many issues still remain. IMO hit detection is extremely broken currently, there is some major issues there which needs fixing. IMO a really good way to show the problem is to shoot a gliding lerk as it flies by quickly... You can start landing hits correctly and then randomly all your shots start missing, even when your tracking it perfectly. You can test this on a lerk flying in a curved line and its usually pretty reproduceable.
Now, even if it's just perception, you still have to address that, or else players global experience is diminished. You have to address it differently then if there was a "real" problem, but it still needs to be addressed. Perception is one of the more important aspects of your global enjoyment of games.
Wathever magic works (or doesn't) behind the code-curtain, all you know is what you perceive.
And I definitely feel that sometimes, the game doesn't play "fairly", in the sense that I am getting hit after being in an "untouchable" situation (behind cover or gone).
With that said, it is understandable that no game can be perfect when it comes to lag/timing/syncronisation and perception. But, I do feel that there is room for improvement. I hope so anyways!
*Btw, when this happens, I'm playing with a sub 75 ping.<!--QuoteEnd--></div><!--QuoteEEnd-->
another thing to remember on top of all this netcode discussion is where the camera is in the Skulk's model
this can lead to you thinking you're in cover when your front claws are peeking out a vent, or being shot around a corner when your back feet haven't passed the corner yet
There was a mod to correct this, Unlagged (http://www.ra.is/unlagged/). It modernized the Quake 3 netcode in many ways, and I believe it predated the implementation of latency correction in most other games. It did what has been described in previous posts in this thread; it rolled back player positions based on the firing player's latency, and THEN did the hit detection.
I was involved a bit in the testing of the mod, and I remember having discussions with the author about latency correction on projectile weapons. Hitscan weapons (instant-hit basically) is easy, but what about when you fire a very rapidly moving projectile like a rocket? The author stuck latency correction for this into a later build, and we tried it out, but there was a notable side effect; the rocket's path was dependent on the starting position of the player firing it, so the normal latency correction approach didn't work; every other player would see the rocket moving on a different path/direction than the firing player and server would.
The end solution was to start the rocket, in everybody else's perspective, from where it was when the other players "found out" about the rocket. In other words, if I fired a rocket, you would see it appear multiple feet in FRONT of me, rather than appearing at the mouth of my rocket launcher. It was not a great solution, but it was the only one that seemed to work.
I'd actually be interested to know how modern game engines like Spark handle the fast-moving-non-hitscan-projectile problem.
EDIT: It looks like later versions of unlagged went with a hybrid approach where he basically bypassed the 50ms of interpolation in certain cases for missiles, letting him get away with up to 50ms of "side-effect free" latency correction on missiles.
well, most popular games have engines derivative of idtech engines. CoD and all source games for example. In Spark, the grenade launcher is an atrocity. It's impossible to aim with it because of the horrible interpolation and the way grenades like to fly into your face because that marine that moved through your view 10 seconds ago counts as being in your way because that's what the server thinks. Thanks UWE I always wanted to spend resources just so I can kill myself and lose my jetpack.
My favourite part of the engine is how an onos with 100 ping can hit you all the way across lava falls (literally, it has happened) and a skulk of similar ping would always be able to bite you before he is in one-shot range of your shotgun. That's some QUALITY netcode.
it is one of those "turn your brain of and click" games. IMO the first one had more depth than this one (or it feels taht way because it was slower, or because I was younger?), but what the hell, if it is being played competitively I may have to check it out.
AAAAnd on topic.
Neo, Ive had that happen with the grenade launcher as well, I thoguht I had hit a wall or something, but that makes more sense!
The grenade launcher is too unpredictable, it makes it not fun to use.
it is one of those "turn your brain of and click" games. IMO the first one had more depth than this one (or it feels taht way because it was slower, or because I was younger?)<!--QuoteEnd--></div><!--QuoteEEnd-->
TFC and QTF are completely different from TF2 in competitive play. My favourite movie is Fanom <a href="http://www.youtube.com/watch?v=ENruOrM01CQ&hd=1" target="_blank">http://www.youtube.com/watch?v=ENruOrM01CQ&hd=1</a> as it shows actual teamplay. Although they aren't the best, I do have videos of my own. They are found on my youtube channel in my signature.
- Marine sees you.
- 50ms - his 50ms internet latency - Server receives packet of marine seeing you.
- 82ms - 32ms based on tick rate - Server is done processing marine seeing you.
- 132ms - your 50ms internet latency - Your computer receives packet of marine seeing you.
(meanwhile marine starts shooting you around here if he wasn't already pre-firing)
- 232ms - 100ms interpolation - Your computer is done waiting for data to interpolate between.
- 292ms - 60ms render time - Your computer can easily take this long to render a frame depending on your settings and frame rate.
- 307ms - 15ms monitor latency - Monitors often have some processing lag. It's smaller on TN displays, but this is common. You now see marine.
- 507ms - 200ms human reaction time
- 515ms - 8ms default mouse frequency
- 531ms - guessing on a 16ms processing time here; I don't think it waits for the frame before sending data to server?
- 581ms - your 50ms latency - Server receives your movement to get around the wall
- 631ms - guessing on 50ms to move out of the way
- 663ms - 32ms server processing time
- 713ms - his 50ms internet latency
- 813ms - 100ms interpolation
- 829ms - guessing again at 16ms processing time
(here I THINK the marine can't hit you anymore even though he still hasn't caught up visually)
I'm just demonstrating how things add up. That's like 629ms of him shooting you. I'm probably wrong on some of that (hard to get the order right, maybe someone can do better), but my point is made I hope. Because skulks move so fast, it accentuates all these latencies. You can do SOME things to reduce the problem, though. Turn off vsync,get a higher FPS, turn off/down frame pre-rendering, stay on servers with low latencies, get a low latency monitor, make sure you're using a high frequency mouse. As far as Unknown Worlds, they can improve client and server performance, reduce the interpolation period (67ms would be better, I think, for this tick rate), up the tick rate once server performance improves.
<!--quoteo(post=2025088:date=Nov 15 2012, 05:12 PM:name=PseudoKnight)--><div class='quotetop'>QUOTE (PseudoKnight @ Nov 15 2012, 05:12 PM) <a href="index.php?act=findpost&pid=2025088"><{POST_SNAPBACK}></a></div><div class='quotemain'><!--quotec-->(approximate best guesses based on what I know)
- Marine sees you.
- 50ms - his 50ms internet latency - Server receives packet of marine seeing you.
- 82ms - 32ms based on tick rate - Server is done processing marine seeing you.
- 132ms - your 50ms internet latency - Your computer receives packet of marine seeing you.
(meanwhile marine starts shooting you around here if he wasn't already pre-firing)
- 232ms - 100ms interpolation - Your computer is done waiting for data to interpolate between.
- 292ms - 60ms render time - Your computer can easily take this long to render a frame depending on your settings and frame rate.
- 307ms - 15ms monitor latency - Monitors often have some processing lag. It's smaller on TN displays, but this is common. You now see marine.
- 507ms - 200ms human reaction time
- 515ms - 8ms default mouse frequency
- 531ms - guessing on a 16ms processing time here; I don't think it waits for the frame before sending data to server?
- 581ms - your 50ms latency - Server receives your movement to get around the wall
- 631ms - guessing on 50ms to move out of the way
- 663ms - 32ms server processing time
- 713ms - his 50ms internet latency
- 813ms - 100ms interpolation
- 829ms - guessing again at 16ms processing time
(here I THINK the marine can't hit you anymore even though he still hasn't caught up visually)
I'm just demonstrating how things add up. That's like 629ms of him shooting you. I'm probably wrong on some of that (hard to get the order right, maybe someone can do better), but my point is made I hope. Because skulks move so fast, it accentuates all these latencies. You can do SOME things to reduce the problem, though. Turn off vsync,get a higher FPS, turn off/down frame pre-rendering, stay on servers with low latencies, get a low latency monitor, make sure you're using a high frequency mouse. As far as Unknown Worlds, they can improve client and server performance, reduce the interpolation period (67ms would be better, I think, for this tick rate), up the tick rate once server performance improves.<!--QuoteEnd--></div><!--QuoteEEnd-->
Mmm... its pretty accurate, actually. There are some diffs, but they are basically a wash.
Its the same in pretty much all lag-compensated games, its just that its not as noticeable in slower-paced, longer ranged games. In NS2, 800 ms allows a celerity/wall-jumping skulk to move 10m, so you can be dead before you start moving. In CoD, 800ms is less than a third of that.
The reason you don't experience it as an 1s delay is that the 1s is that its how long time it takes for a targets reaction to propagate to the attackers world. Get a surprise attack on a marine and try to measure how long time it takes for that marine to start dodging your attack; it will be pretty close to 0.75-1sec or so.
As a marine only needs .4 second of rifle fire to kill a skulk, any skulk waiting until he gets hit before starting to move is already dead.
Which is why good players never stops moving.
But this is all if you're not doing what you should be doing if you can help it -- get behind them.
<!--quoteo(post=0:date=:name=Metal Handkerchief)--><div class='quotetop'>QUOTE (Metal Handkerchief)</div><div class='quotemain'><!--quotec-->... and it really grinds my gears that I can be smoked by someone who visually has their back towards me ...<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->Plays off the quote above, and to me, is unacceptable. My play style relies a lot on predicting player moment and patterns, as it should be for just about any player. With this 'delay' or misrepresented model update, movement prediction is almost entirely thrown out the window. And this isn't to mention the teleporting player models ...<!--colorc--></span><!--/colorc-->
<!--quoteo(post=0:date=:name=Swiftspear)--><div class='quotetop'>QUOTE (Swiftspear)</div><div class='quotemain'><!--quotec-->... For the most part we've had backwards progress in gaming in the last several years.<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->I couldn't have said it better myself ... well, maybe a little emphasis "<i><b>A LOT</b> of backwards progress</i>"<!--colorc--></span><!--/colorc-->
<!--quoteo(post=0:date=:name=xDragon)--><div class='quotetop'>QUOTE (xDragon)</div><div class='quotemain'><!--quotec-->... IMO that is the least of the problems currently, the poor movement, collisions, hit detection and low FPS all hold this game back immensly, and while the devs are and have been working on fixing them, many issues still remain. IMO hit detection is extremely broken currently, there is some major issues there which needs fixing. IMO a really good way to show the problem is to shoot a gliding lerk as it flies by quickly... You can start landing hits correctly and then randomly all your shots start missing, even when your tracking it perfectly. You can test this on a lerk flying in a curved line and its usually pretty reproduceable ...<!--QuoteEnd--></div><!--QuoteEEnd--><!--coloro:#2E8B57--><span style="color:#2E8B57"><!--/coloro-->Agreed 100%
I've been playing FPS for about 17 years now, and other genres for much longer. <a href="https://developer.valvesoftware.com/wiki/Lag_compensation" target="_blank"><i>Lag Compensation</i></a> had it's place back in late 1990's and early 2000's when most of the playing field was 33.6 - 56kbit/s (general population had a 150+ ping) and broadband was just making its debut. In regards to the idea itself, it was a great idea, but just like most things these days, it was good in writing and not in practice.
In my opinion, Lag Compensation should be taken out entirely. I feel it has <i>NEVER</i> worked correctly and give the games that use it a pseudo feel to it. This feeling generally stems from my experience of pre-Lag Compensation days or what I like to call true net code, where Lag Compensation was done by you and hit detection was greater than 99%.
According to <a href="http://www.verizonbusiness.com/about/network/latency/" target="_blank">Verizon</a>, average latency times across the US are 45ms or less. In my experience, if you play with that ping, you can use insta-hit weapons without compensating in almost all situations.
In summary, Lag Compensation, IMO, is no longer needed and adds unnecessary complexity to net code; which still (in 14 years now) doesn't work correctly. In combination with misrepresenting model position and prediction timing on the player's part, is an absolute fail solution.<!--colorc--></span><!--/colorc-->
Use the record command in the console it will have almost zero impact on performance.
Then when you play it back you can use a screen capture program to illustrate the issue you speak of when typically playing.
This would be most helpful for you to accomplish, and try to have r_stats on when you do this.
Also i think a key (spacebar?) plays the demo back at actual fps recorded instead of as fast as your machine can render the demo.
Thanks
0_o, I'm calling BS on this one. Unless you have a lucky connection to a major hub, you're going to get something like 20-30ms just getting out of your residential connection.
Here's my connection to a server just over 10 miles away
<a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459555.png" border="0" class="linked-image" /></a>
And two across the country for me (I'm in Seattle)
<a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459519.png" border="0" class="linked-image" /></a>
<a href="http://www.pingtest.net" target="_blank"><img src="http://www.pingtest.net/result/72459739.png" border="0" class="linked-image" /></a>
Unless I play only on servers within the Pacific Northwest, there is no way I'm averaging on 45ms.
Also, it looks like the Verizon data is for enterprise (think server farms) connections, not residential. I'd double or even triple those latency values to account for residential connections.
Its the same in pretty much all lag-compensated games, its just that its not as noticeable in slower-paced, longer ranged games. In NS2, 800 ms allows a celerity/wall-jumping skulk to move 10m, so you can be dead before you start moving. In CoD, 800ms is less than a third of that.
The reason you don't experience it as an 1s delay is that the 1s is that its how long time it takes for a targets reaction to propagate to the attackers world. Get a surprise attack on a marine and try to measure how long time it takes for that marine to start dodging your attack; it will be pretty close to 0.75-1sec or so.
As a marine only needs .4 second of rifle fire to kill a skulk, any skulk waiting until he gets hit before starting to move is already dead.
Which is why good players never stops moving.<!--QuoteEnd--></div><!--QuoteEEnd-->
Competitive games require higher tick rates, i believe it's 30 at the moment for NS2? CS:GO has 64 on their competitive match making servers and it's pretty bad to say the least, keep in mind CS:GO is a slower game - in terms of movement speed - than NS2, so you can only imagine how bad 30 ticks is for a fast game like NS2. But I believe 64 ticks would be enough for NS2 since you don't really need to aim at small hit boxes - headshot -, so i'm assuming as performance improves, they'll increase the tick rate, right?
thisssssssssssssssssssssss
I know that UWE has addressed the replay system lacking features, but have they addressed the fact that it literally doesn't work at all?
1) you can't view a demo if it was recorded in a different resolution
2) if I put my graphics settings to high it moves in complete slow motion because it just doesn't run
3) if I try to watch it on the *same* settings that I recorded it, it still skips and lags