Are tyrants killing the game?

[Nux]

Take a look here.

i've just added some nitpicky comments there.

On the thread? I can't see them.

[RAKninja-Decepticon]

it was there, now it is not.

it was him asking about several connection related settings, and asking you if you further tested your "delayed animation/sound" theory.

i suspect he deleted his post after actually reading the rest of the thread.

btw, no input on similarity to lagspike behavior?

just a wild guess, but could it be some effect of unlagged's prediction system?

something like:

player a and player b are approaching each other from opposite directions.

player a has 100 ping, player b has 300.

unlagged tries to predict motion by checking where you were moving the last time it received from your client.

player a sees player b coming, and starts to move out of the way.

likewise with player b

collision

all of a sudden, the latency becomes a big issue as each player is getting/receiving commands at different rates, and unlagged is making all sorts of various assumptions, trying to make up the difference, but the result is a collision that lasts a lot longer, and likely will send you off in a random direction.


kind of like what happens when you lagspike while moving.  for a few instants after you spike, you keep doing whatever it is you were doing before the spike (like moving) till the server notices you have not sent any input for a "while".  when the client catches up, you're in a different location.

it's even noticeable with "small" lagspikes (up to 6-700 ping for 2-3 seconds) and very short movements (such as for placing a buildable in an exact spot)

[Nux]

I've just read through this (had read parts of it before but this time only skipped the part on how to implement it). He mentions the "delayed animation/sound" as a reason for perceived false hits/misses. He talks about how false prediction can give you a bad idea of where the server sees you and how clumsy it can feel when your client is updated with your actual position. Also notable is that the older version of unlagged had buggy interpolation. For the most part he says pretty much exactly what I guessed might be happening. Additionally, I'd expect that having played tremulous for so long, my brain, in an attempt to compensate for counter-intuitive feedback, interprets events incorrectly so that it can process them as if they were intuitive: rendering my observations unreliable.

Yet, I still believe there's something wrong with the calculation of the rollback which is causing a notable difference between the state at which I fired and the state at which it thinks I fired. I'd love to investigate this further, but it would require a hell of a lot of coding in a language I've never used. I'd need to get the server to save the state it rolled back to and the client to save the state it was actually at and then I would be able to see if there was any difference.

According to him the only thing you should have to account for is when to dodge but I'm quite certain I'm also accounting for where I aim which can only be because the rollbacks aren't doing their job correctly.

[ULTRA Random ViruS]

stuff

Take a look here.
Already read that. Just re-read a bit to re confirm my memory.
For me, this hitting behind the target is only noticable for me when the opponent refresh rate is below 10/s approx. particularly when md'ing dretches. I've noticed that either my computer is being a bitch as usual, or that when i shoot middle-front of the dretch while its moving forward, i often miss. Taking demos for you is unreliable as i did say about a year ago about 'demo lag' complaint. If you want some shitloads of examples, just ask and i'll give you some.
All I can say on the matter is the 'aiming behind target and hitting' effect is pronounced and real to me, but not to everyone. I can't be sure why other people I have asked haven't noticed it but am inclined to believe this is because when it does happen, people assume it was they're
 failing and not the failing of the rollback logic. I've noticed a similar problem with melee attacks when aiming in the same direction as you and the enemy are moving (so missing should be out of the question), if you are too close your slash will go beyond them and miss but when you're slightly further away the same attack will hit.
I think 'wrong'. The closest i can relate to them is when they walk 'into' you and you swipe them. It always misses for me, apparently the best strategie againts me as makes my speed go to '0' as well as make me miss and have to turn a full 180.

Another notable frustration I have with tremulous at higher pings (100+) is with collision detection. Yet again I can't be certain that it isn't just the fault of slow feedback (unavoidable with higher latency) but it seems as though beyond having less time to react to collisions, the collisions last for longer and any prepared counter movements are ignored; making the merest touch with an enemy or ally disastrous to your escape. This is quite a pain to me as well, especially when all of the sudden, an ally walks into or in front of you, or even under you and you land on them, and ur speed pretty much slows to 0 immediately or slows down significantly. Even worse is the shove effect, but thats something else.
One thing that annoys me most about unlagged is the time when the computer accepts you've swiped and when you do click. I would personally like to have the unlagged calculations from the previous frame before, not on the current frame when the computer detects it as often this is a pain for me. My crappy computer delay + framerate delay all adds up to about to an additional 100 ping for me. Another annoyance is the sv_fps being so low but thats kinda unrelated.

lagged

WRONG.

What am i wrong about? That you don't need unlagged below that certain ping? Its just my opinion, not sure if you are used to playing lagged. I'm still learning to aim ahead particularly at 80 ping.
I've also won matches on 400+ping servers lagged and i've recorded it LIVE. I was kinda being a bitch halfway though by refilling my mass driver while a marauder snipes my base. [korx mod]

[/dev/humancontroller]

lagged

WRONG.

What am i wrong about?

fact: there is no such terminology "lagged" to denote games that do not employ the features of Unlagged, a Quake 3 mod.

[Nux]

[The] thing that annoys me most about unlagged is the time [difference between] when the computer accepts you've swiped and when you click. I would personally like to have the unlagged calculations [start] from the frame [previous to] when the computer detects it.

You're going to have to explain a lot of what you mean there: When you say 'computer' do you mean client or server? Are you judging the time 'the computer accepts you've swiped' by the time you see blood or hear a sound effect? You do know the unlagged calculations are serverside and take into account things like built-in lag, right?

I don't see what looking at the frame before the 'current' one does to help. If the state estimation is wrong, it's wrong by a lot more than just one frame.

[ULTRA Random ViruS]

fact: there is no such terminology "lagged" to denote games that do not employ the features of Unlagged, a Quake 3 mod.

I just use that term to state 'not lagged'. Its a few characters shorter and convenient to type. Why do people call people who stay in the sky with jetpacks 'jettards' or 'jet campers'?

[1]You're going to have to explain a lot of what you mean there: When you say 'computer' do you mean client or server? Are you judging the time 'the computer accepts you've swiped' by the time you see blood[1.1] or hear a sound effect[1.2]? You do know the unlagged calculations are serverside[1.3] and take into account things like built-in lag, right?

[2]I don't see what looking at the frame before the 'current' one does to help. If the state estimation is wrong, it's wrong by a lot more than just one frame.

[1] Basically i want the unlagged to calculate the calculations from the previous frame sent to the monitor rather than the one comming up. It becomes very noticable than you think with low fps.
[.1]No. Thats crazy. If i say thats true, i'd be saying i'd see 0.5 second lag. You know my everyday ping if you've seen me play. About 95% of my games are NOT in australia because not many australians play trem anymore, particularly on australian servers or local. The closest i can get for recent games is SiR/Asia unofficial server with 200 ping.
[.2]The sound effect is late but i ignore sound because i usually play without sound anyway.
[.3]lolwut? I thought the unlagged calculations are compiled in the cgame file. Server-side is whether the server accepts your requests or not.

[2]AS i state, it does make a massive difference for ME and others on crappy computers. I usually react slightly before the crosshair goes of the dretch, then fire upon the exact or close enough frame [my constant framerate is about 40 on atcs in middle with no one around me with my current settings, and 20 on lowest settings {yes, lowest settings make it WORSE, i do not understand how so dont ask me} then the computer detects/calculates the frame later as a miss which is true, it did miss when the computer saw me shoot. This results in my miss-hit ratio of about 15:1 with the dretch using walls. I had high fps in the old days, but my reaction time was crap yet i managed to hit a dretch once every 5 shots. Often theres only about 1.5 or less frames for me to shoot within for me to shoot. On bigger maps like arachnid, i often have to simply be lucky because if i turn about 30 degrees a second (pretty slow looking for my screen dimensions, which is something like 250:140 or something like that, i cant remember, or to be exact, 720:580 pixels) and the dretch about 300 units away, which is quite average, in the alien base, (gives me 16-8 fps or less without any particals or structures), well i have pretty much less than 1 frame for me to shoot the target. In other words, an almost guaranteed miss. Not suprisingly, in arachnid i get repeated sounds, and even worse, those repeated sounds actually 'tear'.*
_________________

*Try to imagine your computer frozen, but still runninig. Tremulous does things like repeat sounds at that timeframe every second. Then try to imagine that sound screwing up like... you should know what i mean.

EDIT: [2]simplyfied: My framerate is low enough that it makes a big difference for me.

[Nux]

You do know the unlagged calculations are serverside and take into account things like built-in lag, right?

lolwut? I thought the unlagged calculations are compiled in the cgame file. Server-side is whether the server accepts your requests or not.

If the server trusted your client to say whether you hit or not then things like aimbots would be pointless since you could have your client send a 'I HIT EVERYWUN!' message at every given opportunity. Backward reconciliation occurs on the server. The only thing your client does is render what the server said the game looks like and send your desired actions to the server every frame.

So let's say you have a ping of 100ms and fps of 50. There are two different 'what I did' events: The one your client requested at some point in time and the one the server decides upon 5 of your frames later. I think the problem here is that you're trying to describe what you think is happening behind the scenes rather than what you explicitly see happening.

[ULTRA Random ViruS]

If the server trusted your client to say whether you hit or not then things like aimbots would be pointless since you could have your client send a 'I HIT EVERYWUN!' message at every given opportunity. Backward reconciliation occurs on the server. The only thing your client does is render what the server said the game looks like and send your desired actions to the server every frame.

So let's say you have a ping of 100ms and fps of 50. There are two different 'what I did' events: The one your client requested at some point in time and the one the server decides upon 5 of your frames later. I think the problem here is that you're trying to describe what you think is happening behind the scenes rather than what you explicitly see happening.

Aimbots work on pure servers i'm sure, because it uses the standard vm from data-1.1.0.pk3.
If you want a 'i hit everyone' you'll need a server that excepts you modded vm. And the server probably has to be modded as well to do so.

[/dev/humancontroller]

Aimbots work on pure servers i'm sure, because it uses the standard vm from data-1.1.0.pk3.

WRONG.

[Nux]

If the server trusted your client to say whether you hit or not then things like aimbots would be pointless since you could have your client send a 'I HIT EVERYWUN!' message at every given opportunity. Backward reconciliation occurs on the server. The only thing your client does is render what the server said the game looks like and send your desired actions to the server every frame.

So let's say you have a ping of 100ms and fps of 50. There are two different 'what I did' events: The one your client requested at some point in time and the one the server decides upon 5 of your frames later. I think the problem here is that you're trying to describe what you think is happening behind the scenes rather than what you explicitly see happening.

Aimbots work on pure servers i'm sure, because it uses the standard vm from data-1.1.0.pk3.
If you want a 'i hit everyone' you'll need a server that excepts you modded vm. And the server probably has to be modded as well to do so.

Sorry but it's like you're talking to someone else. Where did I say aimbots weren't able to work when you're connected to pure servers? I said that if the server asked the client the question "Did you hit an enemy?" (which would be the case if the unlagged rollbacks occurred on your client) you could just mod your client to answer "Yes. All of them." which would make something that automates your aim redundant.

The fact of the matter is that the server does not trust the client to tell it whether you hit or not. The server will be the judge of that.

[Flux]

The real question is: is the game killing tyrants??

[ULTRA Random ViruS]

All i was trying to say is that the aimbot is built into the .exe, not the actual data.

The real question is: is the game killing tyrants??

Yes and no. To me in gpp, its completely useless unless you have under 250 ping. With an exception of charging all over the place in the human base for a few seconds. But my fps isnt high enough for me to do that, each frame takes up about one turret length displacement on full charge speed on a standard atcs layout.
Tyrant is too much of a hit and run because of its healing speed relative to its 1.1 version, i'm too used to 1.1 rant.
I think luci should be powered down to 2.5 second charge, the devs said it was stronger in one post but it really isn't, it does the same damage. After that the charge should even out the balance.