User panel stuff on forum
  15 posts on 1 page  1
Server Talk
2024-12-27, 05:35
Member
96 posts

Registered:
Dec 2008
Did video with antilag bugs, demo included.
In the movie possible to see how model of shaft hits player, but server decides that its not hit

Description:
- above player is the bar, that shows health and armor state - when getting damage, this bar allow you to see this fact, this is server-side health vision.
- sprite of blood painted only when shaft model hit player model, that client-side-on-server (mvd) vision.
- and you can easy see in this video, that blood appears more often, than healthbar changes, that means real hits does not count by server as real, that why shaft now broken.

Now watch second demo - from 2008 year. There every hit with shaft model decreases healthbar.
That means now server counts not some of hits, bugged.

Videos and demos here: https://t.me/+SyYo7LdNNENiYjQy
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-01-19, 16:02
Member
96 posts

Registered:
Dec 2008
Previously i did investigation about model and blood and hit in mvd (server-side) on antilag server, and found no bugs here.

But now i did record video with Fraps on povdmm4 with antilag but without pent on start.
And then downloaded mvd demo from server to get server vision, and compared with original PC view.

So we can see how same moment look from player side(video) and server-side (mvd).

Found moment when player get first hit with cell, there should be blood.

And in video its showed well.
But in mvd demo blood does not appers.

Strange, that showed shaft model in server-side record directly hit pleyer model.
But antilag decided that its not hit.
Its a bug.

http://zachem.org/temp/Antilag_issue_1_1.jpg

http://zachem.org/temp/Antilag_issue_1_2.jpg

http://zachem.org/temp/Antilag_issue_1_3.jpg


So it means that antilag wonders apperas only when shaft model hits player, not before.
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-01-20, 07:31
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
Previously i did investigation about model and blood and hit in mvd (server-side) on antilag server, and found no bugs here.

But now i did record video with Fraps on povdmm4 with antilag but without pent on start.
And then downloaded mvd demo from server to get server vision, and compared with original PC view.

So we can see how same moment look from player side(video) and server-side (mvd).

Found moment when player get first hit with cell, there should be blood.

And in video its showed well.
But in mvd demo blood does not appers.

Strange, that showed shaft model in server-side record directly hit pleyer model.
But antilag decided that its not hit.
Its a bug.

http://zachem.org/temp/Antilag_issue_1_1.jpg

http://zachem.org/temp/Antilag_issue_1_2.jpg

http://zachem.org/temp/Antilag_issue_1_3.jpg


So it means that antilag wonders apperas only when shaft model hits player, not before.


Hello,

blood is not predicted client side so any blood you see is the server confirming a hit (thus, server side). Where the shaft model is drawn is irrelevant in antilag situations. Also keep in mind that the majority of .mvds are recorded at 30fps (server) unless the server is manually modified to record at 77fps. There's also interpolation in demos and they will not always be accurate in showing what actually happened, and much less so at 30fps.

Maybe I'm misunderstanding, but I hope that helps!
2025-01-30, 20:40
Member
96 posts

Registered:
Dec 2008
Can you tell me then - what alghoritm of antilag in game?

In gamer there is command
cl_fakeshaft 0-1-2

2 says that it is special for antilag servers - what difference between model showing then?
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-01-30, 23:38
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
Can you tell me then - what alghoritm of antilag in game?

In gamer there is command
cl_fakeshaft 0-1-2

2 says that it is special for antilag servers - what difference between model showing then?


cl_fakeshaft is only different versions of client-side animation of how to draw the beam from the thunderbolt weapon. Antilag (all forms of antilag) make it so that you should always aim with your crosshair - the server will check if you're aiming at the enemy or not and then decide if you did damage, which means that cl_fakeshaft 1 is the only "correct" fakeshaft to use if you want to properly display what's going to happen.

cl_fakeshaft 0 will show the old animation of where your cells hit in the past (never use this as help for your aim)
cl_fakeshaft 1 will simply draw the animation immediately, working essentially like an elongated crosshair
cl_fakeshaft 2 is a faster animation of what LAN-ping used to look like pre-fakeshaft. It'll still be "wrong", but many scandinavian players are used to that animation so they still use it.

Neither of these settings affect the server and are only client-side animations.
2025-01-30, 23:43
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
Can you tell me then - what alghoritm of antilag in game?

In gamer there is command
cl_fakeshaft 0-1-2

2 says that it is special for antilag servers - what difference between model showing then?


By the way, I know Discord has been blocked in Russia but if possible (if you use VPN or whatever), it'd be easier to help out if you asked questions in the main QW discord channel.

https://discord.quake.world
2025-01-31, 21:27
Member
96 posts

Registered:
Dec 2008
Thank you for advise, now i will use fakeshaft 1 and fakeshaft_extra_updates 1 too - to get always aimed shaft model on crosshair.

But its very strange - how client can see own action from server - nohow? Its just registered on server without feedback to client?

Cant use discord - dont have vpn.

So if server takes client-view aiming to decide that you hit or not, how much fps sends to server from client then?
It is possible to know how fps on client changes this or its rate dependent or something else?

And how about cl_predict_players and cl_predict_half and cl_nolerp - they changes client-side enemyes model positioning. What sould i set them or there is another reality too, but eachside client reality?

Update: today found one more bug as i think - shaft model twice going through enemy and dont count as a hit.
Even if mvd demo records 30 fps, then how it can be possible - dont count this hit with 77 fps from client?

Video here: ftp://blaze:blaze@qw2.ru/ezquake%203%202025-02-02%2019-19-05-41.avi
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-02-02, 20:01
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
Thank you for advise, now i will use fakeshaft 1 and fakeshaft_extra_updates 1 too - to get always aimed shaft model on crosshair.

But its very strange - how client can see own action from server - nohow? Its just registered on server without feedback to client?

Cant use discord - dont have vpn.

So if server takes client-view aiming to decide that you hit or not, how much fps sends to server from client then?
It is possible to know how fps on client changes this or its rate dependent or something else?

And how about cl_predict_players and cl_predict_half and cl_nolerp - they changes client-side enemyes model positioning. What sould i set them or there is another reality too, but eachside client reality?

Update: today found one more bug as i think - shaft model twice going through enemy and dont count as a hit.
Even if mvd demo records 30 fps, then how it can be possible - dont count this hit with 77 fps from client?

Video here: ftp://blaze:blaze@qw2.ru/ezquake%203%202025-02-02%2019-19-05-41.avi


The way antilag works is that every time you shoot client side, once the server receives the information it'll rewind time, check if you hit, then resume time. It's not actually the client telling the server if it hit or not, but the server rewinding time to check for itself. Remember it's server authorative and clients can't be trusted.

The way antilag works means demos will look weird at times. One example would be if you record a .mvd (server view) of a high ping player (or better yet, just spectate one on qtv), the shaft beam will not line up with where the damage was dealt. The beam will lag behind considerably but damage will still be dealt, since the server has calculated that the shooter actually aimed correctly. This also means that the high ping player could aim far ahead of the target (and miss) and the beam would "seem" to be on the enemy but won't actually connect with any damage.

cl_predict_players should be 1
cl_predict_half should be 0
cl_nolerp should be 0

These don't affect antilag, only how smooth models will move on your client side view.
2025-02-02, 20:18
Member
216 posts

Registered:
Feb 2011
Spzman wrote:
Even if mvd demo records 30 fps


In case you're not aware, mvd can record at 77 fps so you will get every server frame in the demo instead of interpolating during playback. In mvdsv.cfg:

sv_demofps 77 // record demos at 77fps instead of the default 30fps
2025-02-03, 00:04
Administrator
654 posts

Registered:
Nov 2008
BLooD_DoG(D_P) wrote:
Spzman wrote:
Even if mvd demo records 30 fps


In case you're not aware, mvd can record at 77 fps so you will get every server frame in the demo instead of interpolating during playback. In mvdsv.cfg:

sv_demofps 77 // record demos at 77fps instead of the default 30fps


^ This can help, but will still not be accurate with thunderbolt beam when antilag is involved, so can't really look at it and judge it by the animation. Only works on 0 ping
2025-02-05, 15:52
Member
96 posts

Registered:
Dec 2008
Andeh wrote:

The way antilag works is that every time you shoot client side, once the server receives the information it'll rewind time, check if you hit, then resume time. It's not actually the client telling the server if it hit or not, but the server rewinding time to check for itself. Remember it's server authorative and clients can't be trusted.

This basically means that shaft make hit not as i did - its another verctor and point of aim, and that why player model dont move as it shoud be. It means antilag changes game physics, and what i and some players feels.

May i ask who gaved right to dev's destroy game mechanics without asking of most players, or here we have dev's monopoly or just dictature?

And why server settings now record only 30 fps by default?
Maybe this should be an option - to smooth (and loose) real game expression of players, disk economy it is a bullshit, who needed that?
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-02-06, 01:32
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
Andeh wrote:

The way antilag works is that every time you shoot client side, once the server receives the information it'll rewind time, check if you hit, then resume time. It's not actually the client telling the server if it hit or not, but the server rewinding time to check for itself. Remember it's server authorative and clients can't be trusted.

This basically means that shaft make hit not as i did - its another verctor and point of aim, and that why player model dont move as it shoud be. It means antilag changes game physics, and what i and some players feels.

May i ask who gaved right to dev's destroy game mechanics without asking of most players, or here we have dev's monopoly or just dictature?

And why server settings now record only 30 fps by default?
Maybe this should be an option - to smooth (and loose) real game expression of players, disk economy it is a bullshit, who needed that?


To some degree it changes the mechanics - instead of aiming "in front" of enemies with LG you should aim directly at them with the crosshair (and all other hitscan weapons like SG and SSG).

Now, about development - development began in 2008 of the first antilag and then in 2020 or so with the second antilag iteration. Developers continue developing the game, however all antilag-related features can be disabled by voting `/antilag`. So you have the option to play without it if you'd like, just like you can play without client side predictions by using `cl_nopred 1`.

You shouldn't flame developers for continuing development of the game ;-)

Let me also add that server fps is still 77 as it's been for many many years (up from 72 at some point many many years ago or so). It's only MVD recordings that were default 30 fps for the longest time. nQuake has updated its nquakesv defaults to 77 fps though, I think. But again, it's only for demos. Server fps remains 77 as it's been forever.
2025-02-11, 06:46
Member
96 posts

Registered:
Dec 2008
I understud - now quake have no server reality at all.

Its very sad - developers killed real e-sport in game without asking anyone.
Fake reality is not e-sport, because now actions of players dont affect eachother directly and in realtime, like it was before antilag age.

Now its schizophrenia (devided reality) where you do something and its "corrected" by server - pure idiocy.

Dev's wanted play well on any ping and how they solved that problem - they just broke game, great!
Dont see any reasons to play seriosly in quakeworld further.

Thank you for your attention.
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
2025-02-11, 18:24
Administrator
654 posts

Registered:
Nov 2008
Spzman wrote:
I understud - now quake have no server reality at all.

Its very sad - developers killed real e-sport in game without asking anyone.
Fake reality is not e-sport, because now actions of players dont affect eachother directly and in realtime, like it was before antilag age.

Now its schizophrenia (devided reality) where you do something and its "corrected" by server - pure idiocy.

Dev's wanted play well on any ping and how they solved that problem - they just broke game, great!
Dont see any reasons to play seriosly in quakeworld further.

Thank you for your attention.


The entire point of QuakeWorld as a game is to create three timelines
    Player POV
    Server POV
    Enemy POV


This creates an environment where you can play online with smoothness without having to wait for the server to confirm every action you take. In 1996 when QuakeWorld was released, ID software introduced movement prediction which was groundbreaking at the time in FPS gaming, making online play much more enjoyable (you can turn this off by using `cl_nopred 1`, but it's like ice skating in NetQuake if you're on any ping higher than 12), but with the consequence that you're no longer on the same timeline as your enemy. What movement prediction does is that whenever you press a movement button, your client will instantly respond by moving in a direction. This is "fake" movement as you have not moved yet server-side, but the prediction created by your client is imperative for smooth gameplay. This is one of many reasons NetQuake sucks and QuakeWorld rules.

If your client is on 100 ping with movement prediction, your position client-side will be 100ms in front of where you are server side. This is why on higher ping, you will seemingly "die behind corners" even in vanilla QW. This is due to the fact that the "real" you - the server-side you - is still visible around the corner you're trying to escape, but due to movement prediction you think you have passed the corner.

This effect is further increased the more ping you have.

Antilag works in a similar way. It just lets the server align your ping to whatever would have happened if you were on LAN ping.
If you're looking for "real reality" instead of "fake reality", NetQuake is the only option.

While the introduction of "antilag" (which is really just another ping compensation, antilag is weird terminology) in 2008-2009 wasn't great in how it was handled, antilag is absolutely necessary to be able to play with or against players from outside your own inner region. It's an amazing feat of engineering that is vital to any form of online play in today's gaming universe, which is also why every game uses some kind of ping compensation. Quake was the first game to do it when they introduced QuakeWorld in December of 1996.
2025-02-13, 18:33
Member
96 posts

Registered:
Dec 2008
For thoose people, who like me feels madness of antilag, i found settings, that minimizes antilag distortion.

cl_earlypackets 1
cl_predict_players 0
cl_predict_half 0
cl_nolerp 0
cl_maxfps 154

Monitor frequency must be 77 hz
http://qw2.ru - my servers and demos collection since 1999 via ftp :>
  15 posts on 1 page  1