Group: Member
Location: Finland
No presentation has been written.
Misc  /  29 Apr 2010, 10:35
Antilag explained
Here I'll try to explain how antilag works to the best of my ability and try to answer any questions people might have. I will update/expand this post when I see the need for it and when I have the time.
Changes in antilag implementation

The version of antilag that was on the servers for the last six months had a lot of bugs and issues that have now been fixed in a new version. This new version has been in use since 29.4. on wargamez.dk, qw.intarweb.dk, troopers.fi and quakeworld.fi. Other servers might still have the old buggy antilag implementation.

So please do not confuse the past issues with the current ones.

What is antilag?

Antilag is a server side feature that compensates for lag when players are shooting with instant hit weapons (SG, SGG, LG). Other weapons are not affected, movement or anything else is not affected.

Without antilag a hitbox is used that is not in the place where the attacker saw the enemy when he made a shot. This means that you need to predict your opponent's movement and aim some amount "ahead" of him depending on your ping.

When antilag is used the server notes how much the attacker's view of the enemy was lagged and uses an "old" hitbox. This means that regardless of the ping the attacker can aim directly at the enemy he sees on his screen.

The trade-off with antilag is that from other players point of view the attacker will be shooting to a location where the enemy previously was, not where he currently is.

The reason antilag is a good idea is that the trouble and change in how you need to play the game when you have to predict your opponents movement even a little bit is big. While the trouble and change in how you need to play the game when you see other player shoot to a location where the you or or other player previously was is much smaller.

A picture is worth a thousand words

Here is a picture that tries to demonstrate how things look from different perspectives. There are three black boxes, they signify the position of the person getting shot at (victim) as seen by different computers when they get the knowledge about a shot being fired.

The bottom black box is the position of the victim on the attacker's screen when he makes the shot.

The middle black box is the position of the victim as the server sees it when attacker's shooting command reaches the server.

The top black box is the position of the victim as he himself sees it, when the server's message about a shot being fired reaches him.

The bottom gray box is where the attacker's client would draw the victim at if there would be no prediction of other players.

In this specific example the victim has 38 ms ping and is moving to the right with the speed of 320 ups (normal strafing speed). The attacker has 64 ms ping.

Without antilag the hitbox used is the one in the middle.
With antilag the bottom black box is used instead.
In the old implementation of antilag (used up until 29.4., might still be on some servers) the bottom gray box was used.

The size of the boxes is 32 u * 56 u, the actual hitbox size in QW.

http://kirjasto.org/medar/tmp/antilag_views.png


The letters signify different distances needed to create this picture:

A: The amount client predicts opponent movement. 13 ms with 13 ms ping and 20
ms with higher pings. Predicted time * Victim speed.

In this example: 0.02 s * 320 ups = 6.4 u.

B: Difference between where attacker saw the enemy on his screen when shooting and where server sees him when the command gets there. (Attacker ping - Predicted time) * Victim speed.

In this example: (0.064 s - 0.02 s) * 320 ups = 14.08 u.

C: The amount that the victim has predicted it's own position compared to the server once he gets the information about the shot. Victim ping / 2 * Victim speed.

In this example: 0.038 s / 2 * 320 ups = 6.08 u.

Given these formulas, you should be able to create a picture with any pings and any victim speed that you like. As an another example you can look at a reversed situation where 38 ms player is firing at 64 ms player who is strafing to the right with the same speed of 320 ups.

F.A.Q.

Is hitting with 13 ms with antilag easier than without antilag?

No. If you create a picture like the one above for 13 ms, you will see that the bottom black box and the middle black box end up in the same place. With the old implementation of antilag it was easier to hit with antilag enabled, because the gray box is not at the same position as the middle black box even with 13 ms pings.

If the enemy is in my crosshair will I always hit?

No. There are two things that might make you miss. Prediction of your own position and prediction of the enemy's position. Either of these can be wrong if something unexpected happens (you get hit by a rocket, but don't know it yet). Your own position is predicted by half of your ping, so by 32 ms with 64 ms ping. Opponent movement is always predicted by 20 ms, except if you have 13 ms ping which means that opponent movement is only predicted by that 13 ms.

How does this affect knockback?

The time when you knocked or the amount you get knocked will NOT change because of antilag.

Direction of the knockback is the direction in which the shot was made. So let's have an example, like in the above picture a player is strafing to the right with 320 ups, his opponent has 64 ms and shoots him to the center of the hitbox with LG from 400 units away (LG length is 600 units). Now the difference between what happens with or without antilag can be seen on this picture (bit whacky size, but I wanted to keep it in scale with the earlier picture). Top left red point is where the hit is made with antilag, top right is hit made without antilag and bottom point is where the shot is made from. B is the same value as in above example. A is the firing distance. Alpha is the angle difference, and it can be calculated as arcsin(B/A) * (180/pi). In this example it ends up being about 2.0 degrees and thus hardly noticeable. The value gets bigger when you increase attacker's ping, victim's movement speed or decrease the attack distance.

Does this mean that I can get shot around a corner?

This depends on how you think about it. If you hold onto the notion that what the server sees is the "reality" of the game, then yes in that view you can be shot after you went behind the corner. This is quite on arbitrary way to think about it though. In fact the server's view on things is the least important, because none of the actual players involved see it. If you die "behind a corner" it means that on your opponents view when he was making the shot you weren't behind the corner.

The other part to think about is how big is the change. By looking at the example above you can see that with 64 ms ping the difference is 44 ms in time and 14 units in distance when strafing full speed. These are really small numbers and should be really hard to notice in practice.

What about the shaft beam with cl_fakeshaft 0?

SVN version of EzQuake now has support for emulating "13 ms cl_fakeshaft 0" even with higher pings. This should allow people that prefer to base their aim on the shaft beam continue to do so, even on higher pings. The settings in question are:

cl_fakeshaft_extra_updates 0: This makes fakeshaft no longer update every frame, but instead just when a new beam is received from the server. In other words it becomes jumpy like with cl_fakeshaft 0 and allows you to count the numbers of shots made.

cl_fakeshaft 2: This settings is just like cl_fakeshaft 1, except that it doesn't fake the shaft to your current view, but to the view 13 ms ago. So now a valid cl_fakeshaft value is anything between 0 and 1 like before, plus 2 for this new settings.

So with cl_fakeshaft 2 and cl_fakeshaft_extra_updates 0 you should get the same experience with higher pings that you get with cl_fakeshaft 0 on 13 ms.

Test build for Windows is available here.

It doesn't only contain this change, but all the other changes in the EzQuake development version. So keep that in mind. Bug reports are of course welcome.
Comments
2010-04-29, 11:02
God job Medar
Let people know the TRUTH!
2010-04-29, 11:02
*good ofc
2010-04-29, 11:08
I got a feeling I'm a grey-box shafter ;(
2010-04-29, 11:36
So what exactly is the point? Are you making a point for using the new (or old) antilag or against it?

Thanks for the explanation but I had to read through it a number of times. I think some essential information is necessary

For instance: "Without antilag a hitbox is used that is not in the place where the attacker saw the enemy when he made a shot. This means that you need to shoot some amount ahead of your opponent depending on your ping." In my opinion, it is pretty confusing when you talk about an attacker and an enemy. In my opinion, you should focus your discussion on a specific POINT OF VIEW (client) and explain everything from that perspective consistently. Also, when you say "shoot some amount ahead of your opponent" I think you mean that you should predict the movement of the enemy and shoot at the position where you predict him to be in the near future (temporal), but at the moment it could also mean simply in front of your enemy (spatial). This confusion is even more explicit in the next part where you talk about shooting "behind" your enemy. I am not sure if you mean behind him in a spatial sense or temporal sense?
2010-04-29, 11:41
good explanation.
Imo the main reason for all the antilag-confusion in the forum/comments was the poor initial implementation (Client side), that had more dissavantages than the new one (server side).
Balancing the benefits/badThings with the new antilag, i think its a clear win for antilag.

I say, improve it even more, and more, and make it default.
2010-04-29, 11:44
The point is to try to explain the facts about the antilag feature as well as I possible can.

It should be pretty obvious that I'm in favor of the "new" antilag feature, so my commentary can be biased towards that.

And yeah, I'll try to improve the explanation there. It's always temporal.
2010-04-29, 12:05
Antilag changes shafting mechanics a lot more, than you wrote, because shaft drops player back when hit.
When shafting both, antilag does shafting feelless, because im see enemy model elsewhere it must be, when i hit without antilag.

Stop killing game mechanics, its perfect without antilag!
2010-04-29, 12:08
Do you have any idea about the QuakeLive antilag feature? Is it better or worse? You could ask some developer(syncerror etc.) for their point of view. I think they fixed the hitbox errors atleast some patch ago.

Q3cpma cg_xerpclients made it kinda weird with hitbox iirc.
2010-04-29, 14:18
Not sure what Spzman means exactly, but I did add a section about knocback.
2010-04-29, 15:20
Just played some pov, 78ms vs 36, the enemy was teleporting sometimes, and neither of us had PL nor lag spikes, it was on troopers
2010-04-29, 15:29
Also feels like there are less phisics, or lagged phisics which feels like nothing. Like the LG is not pushing that guy anymore. Kinda like in QL, horrible. You say you're doing this in order to help HPB's, but most HPB's (all) i talked to hate antilag, and prefers the original one. Stop doing a dictatorship and thinkg first of who would benefit and at what cost. I feel really bad doing 50% with 70ms, but i feel really nice doing 33%

In my opinion, stop trying to fix it and make it triggable, with all player's votes in a server. We must do a poll or something, because qw is not what YOU, or RENZO, or whoever want, is what community wants.
2010-04-29, 17:40
relax.
this is a good feature, has much more benefits than drawbacks. Besides, it isnt unique to qw.

Edited by mushi on 29 Apr 10 @ 18:41CET
2010-04-29, 17:59
you like it because you have benefit from it, but the rest will sux so i dont need ur bullshit mushi. it fucked up the gameplay, the weapon balance, movement. Install it when it is finished, dont use the community as ur own test animal, it was extremly gay to installing it during the eql/ownage.
2010-04-29, 18:00
too bad Medar have to do the dirty work(explaining it to the community), he wasnt the 1 who installed it to the servers.
2010-04-29, 18:49
Great post by time!, especially I like HPB part (ofc he confused with HPW), well, he confused all what he can confuse and even more.

miku post even more briliant, so deep understanding or the problem and how antilag is working.

Medar, dunno how this all end up, but you either will be cursed or crowned But I alredy can see nimbus over you But think about dark side.. err.
2010-04-29, 18:55
Is antilag part of the reason we've seen some massive scores on e1m2 this EQL season?
2010-04-29, 19:36
qqshake: HPB=High Ping Bastard, I'm not a WANKER, and if you're confused, maybe is my lack of English, or maybe your lack of understanding because we are not machines, we are persons we make mistakes and we aren't perfect. To make things clear:

-Antilag is supossed to help people with high ping.
-People with high ping rejects antilag.

To make things even more clear:

-Dictatorship definition: form of government in which one person or a small group possesses absolute power without effective constitutional limitations.
2010-04-29, 21:05
#16 I had the same thought. If someone cba to investigate this, find demos of both kind of servers and compare sg%.

Of course it's a tactical choice of flooding but antilag will make that tactic more feasible.
2010-04-29, 21:10
It probably had an effect. More precisely it's possible that the old "too powerful" antilag implementation had that effect.
2010-04-29, 22:30
I don't like antilag
2010-04-30, 00:26
same

what about the concern 'killing people around corners''? it should be in the faq imo.

thanks for trying to shed some light.
2010-04-30, 00:34
"...trouble and change in how you need to play the game when you have to predict your opponents movement even a little bit is big."
i totally disagree, anyone that's been playing quake is used to predicting enemys movement. it's called a rocketlauncher.
2010-04-30, 05:06
Time, of course you're going to have a lagged pov. You're on 78ms! Someone on 13ms playing you would not see the same effects at all, and surely a little warping on ridiculous pings is preferable to not being able to play at all. Do you know how impossible it would be to play without antilag on 78ms?

And "all" high pingers that you've talked to hate it? What constitutes a high-pinger, and how many of those are ignorant reactionaries who would hate anything? You do not speak for everyone on a ping above 26ms, and I suggest you stop acting as an elected ambassador just because 2 or 3 people who don't know what current antilag does share your illogical reaction.

Also, there is no such thing as "killing someone around corners". If they die, then they were visible to you when you shot at them. That is nothing but fair. The distance you can move in a matter of milliseconds is not significant when you take into account the speed of human reaction time. This only becomes a noticeable effect on insanely high ping times, and how often do you see those? I'm sure something could be added to the antilag code to cap it at a certain level, but anyone playing with a ping of 1,000ms has bigger problems to worry about.

Deus: "I totally disagree, anyone that's been playing quake is used to predicting enemys movement. it's called a rocket launcher."
This is nonsense. Predicting someone's movements by over 3/4 of an entire player width (as with 78ms) with a supposedly "instant-hit" weapon such as shotgun or lightning gun was never a part of the game, nor is it as simple as aiming ahead of someone with an explosive that deals a huge area of splash damage; It was a punishment for not living in Scandinavia. It's virtually impossible given the high speed of the quake player and the high skill-level of players these days.

People on high ping are still at a severe disadvantage on this current implementation of antilag, but now they can actually compete. Are people really rejecting this out of hand without exploring it to any reasonable degree? All these nightmare scenarios are based on figments of your own imagination and not facts as to how it actually effects the game.
2010-04-30, 06:06
Squeeze is the ultimate proof that we dont need antilag...
2010-04-30, 06:34
I play with high ping, even with 50-80ms, as long as the server is stable and the ping and pl dont jump up and down...(can give examples of that from various csn eql games) You just need to adjust to the higher ping... So just because im normally at 26-39ms thats not the reason for not wanting antilag. However, i think i could change my opinion as long as the work progresses and the issues affecting the game negatively are reduced.

It was total bullshit to sneak it in during ongoing eql/ownage seasons tho, please dont fuck up the community like that more times. We dont need extra help in diminishing the number of players. Or will antilag along with new maps and other tweaks bring the holy grail that is NEW PLAYERS OMG! :/
2010-04-30, 06:35
Except Squeeze, at his current level, is easily one of the highest-skill 4on4 players there ever was in North America and he's languishing at the bottom end of the division 1 skill barrier at best solely due to his ping.

Not to mention the INCREDIBLE levels of whine whenever someone who is used to playing on 12ms is forced to play on something higher. Imagine how much easier Locktar vs Avenger at the end of last season's Ownage would have been.
2010-04-30, 06:38
Also, you can't "adjust" to hitting good lg or shotgun with high ping. It's impossible to do with any degree of consistency due to the nature of the game, and those are the only things that antilag effects.
2010-04-30, 06:41
I think its lame to whine about high ping as long as the server is good. There is a difference between high ping on a stable server and high jumpy ping and pl .
But im focusing on 4on4... you can do different tasks in the team depending on how high/low your ping is so thats easier to adjust to than in a duel where you have to do everything yourself
2010-04-30, 06:50
"Squeeze is the ultimate proof that we dont need antilag"

I think Squeeze has only been able to compete at all _because_ of antilag. Many people have been wondering how he has been able to hit. Now we know.
2010-04-30, 06:54
There's a difference between accepting that you have to play on 50ms and actually liking it.

You're saying that if you can't use lg or shotgun with high ping then don't? Why should you have to avoid using them? It's an unnecessary handicap based off nothing but geography, and the difference between 30% sg and 50% sg decides a lot of 4on4 games. You can't simply decide not to use hitscan weapons when you're battling for dm3 rl, or you just stole an unattended quad as a boomsticker or a shafter.

You're still at a significant disadvantage with antilag's current implementation. You're never going to get the first shot off when you meet someone head-on, and nails, pines and rockets are still lagged, but you're not completely crippled when it comes to hitscan weapons. The game-altering issues are practically non-existent now and the benefits far outweigh the mostly imagined problems presented in this and other threads.
2010-04-30, 07:58
blaze: Squeeze played very good before antilag was turned on.

stev: if its true that antilag does not make the gaming experience worse for those who dont need it then by all means turn it on. but to those responsible: dont sneak in stuff like this more times.
2010-04-30, 08:01
Is it only me who thought this antilag business was part of qw from the start? predict? i always shoot straight at where the enemy is and always has in qw with directhit weapons..
2010-04-30, 08:01
well blAze he's played alot of games outside of those antilag servers with pretty much the same result so even if antilag was the key in him improving he's still proved lots of times that he can perform at a high level with such ping.
2010-04-30, 08:04
Even if this is a small issue i dont see why ppl with no need for antilag should have to adjust to this:

"While the trouble and change in how you need to play the game when you see other player shoot to a location where the you or or other player previously was is much smaller."

The problem with high ping, i mean when its impossible to find a fair server, is never the lowping players problem imo.
2010-04-30, 08:30
But it doesn't change how you play at all on 13ms? Certainly not in the current implementation.
2010-04-30, 08:49
Well valla & hoora, I suppose we just have to agree to disagree on that one then. Of course "good" is very relative term and hooraytio plays in different division altogether. Also I'm not saying Squeeze is a bad player, but that he plays like one plays with 125ms, if antilag is disabled.
2010-04-30, 09:04
2010-04-30, 09:08
Whats the point of bringing individuals into the discussion anyway? As far as I remember squeeze performed solid in fusion before there was any antilag at all, and the reason for later improvement is more due to Fusions activity and improved teamplay when Valla joined, than the antilag feature. Too bad it doesn't look we will see more HPW awesomeness in Fusion in the future..

In any case this feature could be really great to make more people from far away willing to join European leagues, which is entirely positive. If the issues get fixed, this can be one of the best change the scene has ever seen.
2010-04-30, 09:10
First of all Hooraytio you are forgetting about the case that should be most common: Both players/both teams have the same ping. In that case antilag doesn't "punish" anyone, but creates better playing conditions for everyone.

If for some reason you are playing with a low ping vs. a higher pinging player, then yes this feature does affect you as the low pinger by changing the way you see the high pinging player's shaft beam when he's shooting at you. This however is not a huge game changing crippling side-effect, it's very far from it. I suggest that you try it out for yourself with some reasonable pings.

You will still always hold significant advantage when playing with a lower ping vs. a higher ping. High pinging player doesn't get an advantage, he gets less of disadvantage.
2010-04-30, 09:15
Ok, Im just curious as to why the lowping player should have his game changed at all no matter how small it is. Ill try it later and see what it looks like...

Saying stuff like: you still have an advantage doesnt really help. Hey lets cut off your hand, we cut both hands of the other guy so youre still have an advantage. drastic but it describes the problem with arguing like that (antilag is no way near cutting of body parts tho)
2010-04-30, 09:19
Stop reacting blindly and actually play with it, as you have been doing unknowingly since December, and tell us how it affects the way you play in any reasonable way. And don't do something retarded like pinging up to 500ms; Play someone with 50ms and see how it goes.
2010-04-30, 09:23
Well cutting of the hands of the players would make both players' gaming experience significantly worse, so I'm against that. Antilag on the other hand (lol, hand, get it?) makes that experience significantly better and that's why I think it's well worth having.
2010-04-30, 09:24
I don't have a problem with the feature. It allows me to shaft incredibly even when my ping is shit and jumping. I don't think I have suffered much from any weird side-effects.
2010-04-30, 09:37
One thing i have noticed is that players with high ping seem to get very good sg%/lg% (better than they have ever had even on low ping) and that the amount of frags on boomstick-intense maps such as e1m2 seems to have rissen.
2010-04-30, 09:41
That may have been an issue with the previous implementation of antilag. Current implementation matches equivalent 13ms gameplay more perfectly. I have noticed my lg accuracy drop SIGNIFICANTLY since it was applied.
2010-04-30, 09:43
This has been repeated many many times, but here we go again:

With the "old" antilag implementation that was in use for the past 6 months it WAS easier to hit better SG/LG with any ping that it ever was with 13 ms ping without antilag.

This was fixed recently and is no longer the case. Some servers might still have an old antilag version, but wargamez.dk, troopers.fi, quakeworld.fi and qw.intarweb.dk have all been updated. You can use the /about command to see when the MVDSV build in use was made.
2010-04-30, 10:13
good!
2010-04-30, 10:27
I was testing this feature last night with Qqshka for hours, trying to get some differences between 13ms, antilag with prediction fix and without it. It does indeed show results where antilag without the fix might have the best lg%, but it also happens that real 13ms comes second very near to those results. The antilag with fix enabled had the worst scores of them all among the line.

I also blind tested Maga and Rikoll playing pov (they both can confirm the results), switching the settings between the rounds and see how their scores went. These scores somehow confirm my findinds. In short, the current antilag with medar's fix is "as good" as 12ms, or maybe slightly worse, but absolutely not better.

We'll be testing the feature on server with different methods (an aimbot) later today to see some percentages. Qqshka even coded a hitscan weapon that shoots multiple pellets that do not have any spread, but they always hit the same spot. Hopefully this way we can prove something, then check the code once again if something strange comes up.

Edited by Renzo on 30 Apr 10 @ 11:29CET
2010-04-30, 12:32
Aimbot!

Sounds good but can we have the 'in detail' results from guinea-pig-Maga vs guinea-pig-Rikoll, out of interest?
2010-04-30, 12:44
It's hard to do any meaningful test in one day. Pov results easily vary 10% even with the same settings from round to round. You'd have to accumulate quite a bit of test data to say anything really.
2010-04-30, 13:19
Yes, that is absolutely the hardest part. Especially when there is the human element involved. That's why I'm willing to blindtest more people playing dmm4 especially, so I can gather more results. I would be there to collect the scores, and toggle settings (or keep them for few rounds) and the players wouldn't know which setting was enabled.

Darff wrote:

Sounds good but can we have the 'in detail' results from guinea-pig-Maga vs guinea-pig-Rikoll, out of interest?

Maga played vs rikoll, 9 rounds of pov. I toggled the settings without them knowing, and I was checking Maga's performance. These are the results (I had to dismiss one round, because maga was shot midair all the time, reducing his performance too much).

38ms, antilag enabled, Medar's prediction fix disabled: 2 rounds, Maga had 42-43% lg in both rounds

38ms, antilag enabled, Medar's prediction fix enabled: 2 rounds, Maga had 38% lg in both rounds

13ms, antilag enabled, Medar's prediction fix enabled: 1 round, Maga had 42% lg

13ms, antilag enabled, Medar's prediction fix disabled: 1 round, Maga had 43% lg

13ms, antilag disabled: 2 rounds, Maga had 42% and 43% lg

Percentages are rounded to nearest integral. My own scores are as follows (outposts with Qqshka, lg only)

13ms, antilag on, fix disabled: 53% 49% 50% 53%
13ms, antilag on, fix enabled: 46% 48% 45% 49%
13ms, antilag off: 47% 46% 52% 52%

These results look similar. We played a lot of other rounds too, but the results varied a bit too much to take those into account (hammerv3 is harder map, since it has elevation and hiding places).

This is the reason we need to test it using something that doesn't have the human factor involved, hopefully we can do that later today.
2010-04-30, 13:27
Stev: Do you know how impossible it would be to play without antilag on 78ms?

I know better than no one, but if you prac it, you are able to score up to 35% lg (I did up to 42% before changing my ISP), which is more than decent. The killing behind corners is specially noticeable while speccing, I was speccing Bulat v Matrix, so figure out the pings, nothing more than 78 but speccing it was just horrible. I'm just saying that phisics are really weird, speccing it is horrible and totally unsynched, even with that patch. I know i would be beneficiated with this feature, but i just feel like playing unfair, i prefer shafting with my high ping, no matter how hard is it because the fisics match, because it's harder, and the harder the more fun to me and also because i don't want to change the game's phisics and speccing synch.
2010-04-30, 13:36
time!, physics are not affected in any way. All the hits are treated equally when it comes to physics. What you will probably feel, is some sort of "time lag" when you have been hit.
2010-04-30, 13:40
Interesting, though shame there was no chance to test 38ms with antilag disabled, as that'd tell us quite a bit I think. Ideally you want 30-60ms antilag+fix results vs 13-28ms no antilag results (I'm confident the fix is better than the original from all I've heard). Any chance we could see Rikoll's results for those games too?
2010-04-30, 13:40
I'm sceptical about 35% on 78ms unless you are playing a very bad opponent, or talking about some lucky one time incident, or playing just to save %.
2010-04-30, 13:42
Ask Pericles
2010-04-30, 13:48
Darff wrote:

Interesting, though shame there was no chance to test 38ms with antilag disabled, as that'd tell us quite a bit I think. Ideally you want 30-60ms antilag+fix results vs 13-28ms no antilag results (I'm confident the fix is better than the original from all I've heard). Any chance we could see Rikoll's results for those games too?

Ah yes, we were testing the impact on 13ms only with Qqshka, just to be sure it is not affected in any way.

Rikoll's and Maga's results can be downloaded from troopers.fi servers (ports 28001-3), so you can check them yourselves if you want to. Just cmd dlist maga rikoll pov and you should see 9 demos from yesterday. This way you can also see the non-rounded lg-scores.
2010-04-30, 14:03
If I have to ask Pericles, then you already answered to my question.



j/k
2010-04-30, 14:16
If you're scoring 35% lg on 78ms povdmm4 then you are playing some kind of retarded monkey man. As someone who is part retarded monkey man (on my father's side) even I would be disappointed if I got hit that much against anyone with over 64ms.
2010-04-30, 14:21
#53 Renzo | 30 Apr 2010 @ 16:36 | 88.112.145.26
time!, physics are not affected in any way. All the hits are treated equally when it comes to physics. What you will probably feel, is some sort of "time lag" when you have been hit.

--

Isn't this exactly the kind of negative side affects we do NOT want? And that CANNOT be removed in antilag? As we talked about before, you can NEVER remove lag... all youa re doing is moving the lag from the shooter to to the other player.
Maybe simulating 13ms for ALL ping players is just too drastic and still have too high impact on physics. What about trying what I suggested earlier with ~50% off.

If you have 39 ping it simulate 25, if you have 52 it also simulate 25.. if you have 65-78 it simulates 38.

And then make some fakeshaft 2 variable that actually shows a REAL shaft, lagging with the appropiate ping. so it actually LOOKS like a 25ms shaft or 38ms shaft, so we aren't forced to use fakeshaft 1 and point in middle of player model.

We are not trying to make everyone into aimbots, there should still be skill involved in using shaft and adapting.
2010-04-30, 14:37
i assumed that the lag is still on the shooter - i'm guessing time is saying that when he shafts he doesn't like the delay between his shot and the kickback he sees on his target. the alternative is no antilag where you'd just miss...
2010-04-30, 14:50
No it's the delay between enemy's shot and time!'s kickback, from that shot.
2010-04-30, 14:59
Both are noticeable, but my own kickback is the most annoying, ofc. Enemies' lagged kickback only makes my shafting easier. Also speccing, keep in mind. You see the shaft aiming nowere but in the enemy, and lots blood floating in the middle of the shaft and the enemy. Shaft is in X, enemy is in Y and blood apearing in Z Maybe there's a way to synch that, i don't tryed qtv anyway.
2010-04-30, 16:30
"No it's the delay between enemy's shot and time!'s kickback, from that shot."
Well, if enemy is on low ping, then antilag has no effect on the knockback done to time. Antilagged 13ms knockback happens at an identical time to true 13ms knockback. In fact, even antilagged on 999ms, the time and distance of knockback is unaffected. All that changes is the position of the client's interpretation of the beam with regards to the position that was actually hit. Movement is unaffected, damage is unaffected, knockback is unaffected and delay is unaffected.
2010-04-30, 18:47
I don't play always vs 13ms players
Quote from Arnette after playing with antilag and trying with cl_delay_packet:
arne: easier with 51 than 12
arne:

I don't know if it's old antilag or new to be honest
2010-04-30, 19:37
thank you for adding the part about "hitting players behind corners". i understand you are an advocat of the feature(nothing wrong with that ofc).
here's two comments from my point of view: you mention server reality and the view from the enemy player and explain how antilag plays no adverse role in both scenarios. funny enough you don't mention the most obvious point of view that matters to me. mine. i'm behind the wall when i get shot.
second, you say the time-difference between two players is so short and allows for such a short distance to be traveled that it hardly plays a big role. you assume 'full strafing speed' for this. in quakeworld, the fastest fps i know, 'full strafing speed' almost equals standing still. normal travel speed is easily 2 to 3 times that. top speeds of 5 times that are very common and yes i'm making up these numbers as i type. qw has no limit how fast you can move and the moving speed could very well be 10 times that of strafing speed and you could be 10 times further behind cover than you are calculating it.
2010-04-30, 20:53
Well speedjumping for example, or being shot by shaft you reach pretty high speeds quite fast.
2010-04-30, 21:03
#66

Yes I'm an advocate of this feature, but it's just because I like the feature as a technical solution to a problem and I understand how beneficial it can be. I explain the facts to the best of my knowledge/ability. Antilag is not some agenda that I want to push forward, it's just a technical solution to a problem and I'm trying make people understand why it's a better solution than the one currently in use.

I specifically tried to explain the method I used to come up with all the numbers very clearly and gave the exact formulas that I used, so that everyone can plug in their own scenarios and come up with their own numbers.

You are also very much mistaken about the speeds involved. Full strafing speed is 320 ups, a forward rocket jump is about 750 ups (well that was the best I could do with couple of tests). This is about 2.34 times more than the strafing speed. I would say it's quite rare have any fights where the speeds involved are higher than that. Feel free to test around yourself by using "show speed" in EzQuake.
2010-04-30, 21:13
It's also worth a note, that my example handles the case where someone is moving with 320 ups to a direction that is the normal of the line between the players (ie. has 90 angles with that line). If the movement is in any different angle then the effects will be smaller.
2010-04-30, 21:54
so if you move at rocketjump speed (~750) and say you ping 25 and the other player 65 then you can be fully behind a corner (~32 u) and still be shot. you shouldn't even be visible and you still die, trying to escape with 5 hp. gg.

i don't like this feature.
2010-04-30, 22:13
i can see this being useful maybe in a duel with 2 players that both ping 80 and want to play with instant hitscan weapons. if all players agree that is. anything else or if there's actual ping difference involved i don't see it as an improvement at all.
2010-04-30, 22:36
"you shouldn't even be visible and you still die, trying to escape with 5 hp. gg."
But if your opponent had 13ms then you would have died anyway! The shot still took place before you made this incredibly unlikely manoeuver which may possibly happen once every 500 duels. It's simply justice. To say that an additional cell 65ms earlier would have changed your decision to duck around the corner would be nonsense, so neither your decision nor the result are affected by his large ping and the server's attempt to compensate for it.
2010-04-30, 22:50
what's justice? unequal pings are not justice. no amount of trickery will change that.

to the low pinging player justice is that if he's behind a corner he should be safe. to the high pinging player justice is that he should kill the running player before he can duck behind the corner because 'hey i clicked on him before he was gone!'.
tired, going to bed, don't like feature.
2010-05-01, 00:00
1) Do you have any idea how rare your hypothetical situation will be, and how virtually indistinguishable from normal play it will be when it does occur? How many times has this happened to you in the last 4 months? That's right. I've been aware of this feature for quite some time and I can say with certainty that it has NEVER happened to me.

2) Don't you understand that HE WOULD HAVE DIED ANYWAY SO IT MAKES NO FUCKING DIFFERENCE!?

Play with the feature and judge it on its practical application rather than sit here obsessing over impossible nightmare scenarios that do not effect in-game decisions or results in any way.
2010-05-01, 00:15
The results of testing, with methodology explained.
2010-05-01, 11:09
#70

Here is a picture of the situation you describe. So green guy is is moving at a really high speed (rocket jump speed, 750 ups) while having only a little bit of HP. The red guy has a high ping (64 ms) and is standing in the optimal place so the whole 750 ups speed is moving the enemy behind a corner.

Clearly there are many factors here that make it a very unlikely occurrence. Including the fact that when you trying to shoot someone getting away, you are usually able to position yourself in such a way that the enemy's speed is not at an 90 degree angle compared to the line between the players. The red guy is in the worst possible place he could possible have chosen here.

In any case, when we consider this scenario there is a 44 ms time frame where green guy is completely behind the corner according to the server's view on things. But is partly visible to the red guy. So if he manages to hit a shaft during these 44 ms the player dies while he behind the corner in the server's point of view.

Let's say the red guy is tracking the green guy with 100% shaft accuracy while this movement is taking place. Does this mean that the kill "behind the corner" happens every time? No it doesn't. LG only shoots every 100 ms, so this kill would happen 44% of the time. Also the enemy is not completely visible during this 44 ms time frame. When 22 ms has passed half of the enemy is behind the corner and hitting is hard, when 38 ms has passed hitting is really really hard. But we assume that the red guy can hit 100% even if only a tiniest bit of the enemy is visible.

You got:
- Rocket jumping speed
- Less than 30 health
- Enemy with a high ping
- Enemy in a very weird position to maximize the effect
- Enemy with 100% shaft accuracy
- Enemy being able to hit 100% accuracy even when only a tiny bit of the player is visible

And assuming all that still the changes of this happening are less than 50%.

Then there is also the question about how bad this effect is or isn't...
2010-05-01, 11:12
stev 2) no? without antilag he wouldn't have died in my example.
2010-05-01, 11:28
Made a little edit to the knockback explanation to make it clear that the time the knockback is applied will not change because of antilag.
2010-05-01, 11:46
Deus: He would have on 13ms, which is what antilag is supposed to simulate. If the results are identical to 13ms then the implementation works in this situation. The only reasonable objections are to situations in which it changes the results or the players' decision-making. Saying "antilag makes high ping less of a disadvantage" is not an argument against it.
2010-05-01, 12:03
Medar: i played TF2 and situations like this are common there in my experience. it's not even that my example happens very often. i just absolutely hate how these hitbox-shenannigans give you a feeling about being deceived where stuff really is. it's the opposite to what quakeworld is about: crispness, responsiveness and speed. it gives you some kind of wobbly, vague gameplay where you have to aim next to the player to hit his hitbox and can be shot from around corners.
http://www.youtube.com/watch?v=QkZLJeQ1Nhc
i prefer to play quakeworld the way it is, by much actually.
2010-05-01, 12:04
shakey hands :E
also Stev, yeah well my example had one player with 25 the other with 65 ping.
2010-05-01, 12:10
So? The antilag is still designed to simulate 13ms for all. What I said is perfectly accurate.
2010-05-01, 12:14
I do not know how TF2 netcode works, so I can't comment on what causes the issues you are having there. In any case it has little or nothing to do with QW netcode or antilag.

dEus wrote:
it gives you some kind of wobbly, vague gameplay where you have to aim next to the player to hit his hitbox

But this is EXACTLY the thing that antilag fixes. With antilag you will always have to shoot what you see (with SG, SSG and LG). Not some weird place next to the enemy model.

And just to make it clear: You will never be able to shoot anyone who is behind a corner on your view.
2010-05-01, 18:03
with wobbly and vague i was referring to the problems that come with the distinction between player model and hitbox. it goes against what in my mind sets quakeworld apart from a lot of modern and inferior shooter games.
i know what you're trying to achieve and i also think i understand your reasoning. i also get what you mean by saying that the current qw netcode is just another way of dealing with the problem lag. i just don't agree with your opinion that a new system would bring less problems.
thanks for explaining though, i think it's important to have a clear understanding what the positions in the debate are.
2010-05-02, 10:00
dEus wrote:
with wobbly and vague i was referring to the problems that come with the distinction between player model and hitbox. it goes against what in my mind sets quakeworld apart from a lot of modern and inferior shooter games.

Ok, let's see how things are without antilag:
Shooting with LG with ping 26 or higher: The hitbox is not where the player model is on the screen.
Getting shot by an LG: Your own hitbox is not where your screen is.
Seeing a third person getting shot by an LG (not fired by yourself): The enemy model is not where his hitbox is.

So the only case where player model and the hitbox has ever been in the same spot from any player's point of view is from 13 ms player's pov when he's making the shot. With antilag it's in the correct place for the shooting player regardless of the ping. In the two other cases the difference between model position and hitbox position is increased based on the attacker's ping.

So QW has always had a distinction between player model and the hitbox. With antilag that distinction is removed in the cases where it matters the most.
2010-05-02, 12:31
i'm missing something here or we're talking about different things when saying hitbox. with hitbox i just mean the area where the shot registers for a playermodel. the playermodel and the hitbox are allowed to move forward in the time it takes for your fire signal to reach the server. that's why you have to lead your shots. isn't that right?
but the playermodel and the hitbox are the same, that's why you have to calculate at what time your shot goes off so that it hits the enemy playermodel after your delay has worn of, no? that's how it seemed to work for me for the last few years.
2010-05-02, 15:50
Good news for all the cl_fakeshaft 0 lovers added to the blog post.
2010-05-09, 10:18
Medar wrote:
Shooting with LG with ping 26 or higher: The hitbox is not where the player model is on the screen.

without antilag the hitbox and the model are one - that's why you have to lead your shots. your firing the weapon is simply delayed, e.g. you fire boomstick aiming in front of the enemy playermodel, during the delay the enemy is allowed to move forward and then the shot goes off.

antilag breaks this unity between model and hitbox, that's why i fear it introduces 'wobbly and vague' gameplay.
2010-05-09, 10:42
Stev wrote:
So? The antilag is still designed to simulate 13ms for all.

well, it's impossible. lag is lag and once it's in the equation you can't just take it out. both ways of dealing with lag, new antilag and 'old' antilag (you could say simpy 'lag' ) acknowledge that.
you can't compare my example with a hypothetical 'both players on 13ms' situation. you have to compare the two different scenarios that the two current solutions to the problem lag come up with. in one scenario the player dies (antilag) in the other he doesn't. that was my point anyway.
2010-05-09, 12:48
Quote:
without antilag the hitbox and the model are one - that's why you have to lead your shots. your firing the weapon is simply delayed, e.g. you fire boomstick aiming in front of the enemy playermodel, during the delay the enemy is allowed to move forward and then the shot goes off.


I think that's incorrect because player prediction has already detached the playermodel from the hitbox, even without antilag. Your client already draws the playermodel in front of the actual hitbox, so actually you need to aim at the model, not in front of it.
2010-05-09, 13:23
why is there a delay then when you fire your weapon and you're lagged? it increases with your ping, too.
isn't that delay what compensates for lag while the new antilag compensates for it with the detachment you're describing?

you don't aim at the model with the old lag settings you aim in front it. you predict enemies movement. that's exactly what new anti-lag seeks to 'fix', no?
2010-05-09, 14:02
I don't think I understand what you mean. If there is a noticeable delay in firing a weapon, isn't that simply a result of lag, not any kind of compensation for it. As far as I know, previously the compensation for lag has been the player movement prediction. Unfortunately that has worked properly only for somewhat low pings 13-25. Antilag adds another type of compensation so that you can aim at the playermodel with not only 13-25, but also above.

I have always aimed at the model and have got reasonably good results with both lg and sg, but I have played mostly with pings where the player prediction works as intended.
2010-05-09, 14:07
lg and sg are both instant hit weapons, you are not supposed to make any kind of predictions with them. It's just point and click. It's the projectile weapons that you are supposed make prediction shots with.
2010-05-09, 14:15
yes, true. it's just that if there's lag even your hitscan weapons have to be handled like projectile weapons. it's done like that to achieve the 'unity' i was referring to: if you have ping 100 to a server and you click boomstick attack it will count to 100ms before your shot actually fires 'on the server'. that way there is only one 'reality' - the 'server reality'.

it makes sure model and hitbox are always the same and it comes with the disadvantage of having to wait for your signal to travel to the server before your shot goes off.
2010-05-09, 16:07
Like Medar explained, because of player prediction there has never been only one "reality", but instead there has always been 3 different points of view: client a, client b and the server. Even before antilag, the clients have detached player model from the hitbox and predicted it forward 13-20ms to compensate for the lag. I guess you could say that up until now, only client side lag compensation has been done. With antilag we complement the client side lag compensation with server side lag compensation and combined get a better overall experience. If you really want to have your player model and hitbox the same, you need to use cl_predict_players 0 I think.
2010-05-09, 16:13
In other words it's false to think there has been any unity: there has always been compensation for lag, but it has only worked as intended for 13-25ms. So not only do you have the ping advantage itself, but also the lag compensation mechanisms only work properly for you, if you had low ping.
2010-05-09, 17:15
i didn't know about the cl_predict_players command also i can't find where Medar supposedly is talking about different points of view in pre-anti-lag implementation.
i don't really care if there always has been a distinction between model and hitbox. i mean i'm sure there was, they're different things after all.
what i care about is that there is this delay between you clicking sg +attack and waiting 100ms until the shot is fired on the server. right now you would have to predict where the enemy's player model will be in 100ms and click there. so when your shot goes off it hits the playermodel and also the hitbox which are (more or less maybe, i'm not 100% sure) in the same place.
that's how quakeworld works up to now. i'm speaking from experience, playing teamfortress sniper with 200+ ping gives you a pretty good idea of how the mechanic works.
if you want to make hitscan weapons 'go off' instantly from the firing players point of view regardless of ping and movement speed you would have to introduce a much more significant distinction between model and hitbox and that's what i fear will screw up the feel of the game for me and possibly some others.
that's what i mean with 'unity' vs 'wobbly and vague'. i mean, even if there always has been a (technical) distinction between model and hitbox, why is there the delay in firing your hitscan weapons in the old/current implementation?
2010-05-10, 02:14
Quote:
why is there the delay in firing your hitscan weapons in the old/current implementation?

Info about effects and projectiles comes from server, so you have to wait at more than your pingtime (due to additional delays of client and server processing) to server to make it visible on your client.
2010-05-10, 07:05
Quote:
what i care about is that there is this delay between you clicking sg +attack and waiting 100ms until the shot is fired on the server.


I don't think this is changed in any way by antilag. It will still wait a response from the server.
2010-05-10, 17:40
yes you're right. the delay between click and firing of your weapon doesn't seem to be affected. it still increases with your ping, too. couldn't test it as much as i like though as delay_packet doesn't support high enough values for me to tell for sure if the delay is effected in any way by antilag. is it?

that's a major concern out of the way for me. i'm still against it though just on the basis that i never noticed any discrepancy between model and hitbox with the old implementation and i'd rather not see any changes to fundamental things like this.
2010-05-11, 07:14
I don't think antilag really changes anything in that relation. It will simply use the hitbox that was attached to the model when you shot at it. The way I see it is that the hitbox and model are kept together better with antilag, than without it. With antilag, if you shoot at the model, you will hit the hitbox. This is the way hitscan weapons are supposed to work. Without antilag you shoot off the model and may or may not hit at the hitbox.
2010-05-11, 07:20
To summarize gameplay with lag:

With antilag:
you shoot at the model -> you hit the hitbox
you shoot off the model -> you miss the hitbox

Without antilag:
you shoot at the model -> you miss the hitbox
you shoot off the model -> you may hit the hitbox
You have to be logged in to be able to post a comment.
Username:
Password: