CS Bullet Pattern Analysis

[Darklord]

as I understand, the client/server-issue is like a "Possible World Scenario", right? like, when I fire and my client draws decals, they COULD actually hit there, but the server decides where they actually hit. this is just to say, firing away on your own will actually give you some idea of where the bullets COULD go... and if you do it enough times, you'll get an understanding of the spread... right?



correct me if I'm wrong (garry)



:)

[MIKE121]

http://www.summerblue.net/games/cs_report/index.html

[Spoonman]

rgr that.

[Darklord]

yes I know but people tend to think that what you see has nothing to do with what actually happens - which is true - and that firing into a wall won't teach you anything - which is not true... that's all I was getting at...



wish the summerblue-dudes would update their report... I like it.

[garry]

cl_lc 1 = lag compensation ON.



This means that it will predict VISUAL things on the client side. This is all it means. You can turn this off by using cl_lc 0, which waits for the server to tell you exactly what happens, so you ill get a slight delay.. Unless of course you're playing on the server or a lan.



If the client dll and server dll were compiled at the same time then the recoil physics should follow the same rules, so although the bullets didnt REALLY land there - if you did it again they COULD have..

[Darklord]

ah the beauty of induction...



one comes to think of David Hume

[joe]

garry wrote..

Yes, I also assume he did these tests on a server, rather than setting up a game on his own computer... You're a fkn genius.

i'm not sure why you are seeming to agree with me (albeit incorrectly), then making a sarcastic insult.



in fact, your reply is incorrect - any client he plays on is stil a CLIENT, even if it happens to be on the same machine. the server process and the client process are two different programs, and they each do their own bullet calculations. so your comment leads me to believe you really don't understand what we are talking about.

[Darklord]

teach us the way of the wise



we should all have entrances like Kramer.

[garry]

Edited by [user="26703"] @ [time="1029450776"]

joe wrote..

i'm not sure why you are seeming to agree with me (albeit incorrectly), then making a sarcastic insult.



in fact, your reply is incorrect - any client he plays on is stil a CLIENT, even if it happens to be on the same machine. the server process and the client process are two different programs, and they each do their own bullet calculations. so your comment leads me to believe you really don't understand what we are talking about.

This is all meaningless. Whether the bullets were predicted by the server OR the client dll is irrelevant.. They both have the same code.. Even if the bullets don't really land there, they COULD have.. since they were produced by the exact same recoil values etc..

[Grocer]

Right. The important part is not where the bullet holes actually appear, it's the relative spreads produced by different guns. The report at Summerblue will clue you into all sorts of shit it's good to know, but it doesn't have individual weapon comparisions, which is what Mazuko did. If you want to see how weapon accuracy/spread has changed over time check out this link. The page is in German, but the pictures are self explanatory.



http://www.ggf-gaming.de/weapons/dc_accuracy.php

[Unregistered]

David Johnston wrote..

Hey everybody, lets petition Valve to stop spending needless hours on polishing TF2/HL2/TFC/Wolfenstein2050AD and PongForever and instead direct all their efforts on producing code based on a year of research into gun recoil and bullet trace paths to give only a marginally realistic game hyper realistic bullet paths!

You don't need a year's research to find out that every bullet will hit near to the last forming a line of hits rather than a scattered area.



The random element to recoil is indicative of a game programmed by one, amateur coder (gooseman). He used the system for ease but it has never been updated in any patch as it is so integral to the game.



But with CS2 just maybe things will be different...

[//z38stinger]

bleh! i dont need a link on "bullet locations". i got skill :P. WELCOME TO THE NATIONALISTS DICE! ISNT IT WONDERFUL??

[-Dice]

Power...Ultimate Power!

[LuftWaffle]

Dearest Garry,



The bug where bullets didn't land in the same place on the client and server was a result of not sharing the random seed. They fixed this problem by.... you guessed it, sharing the random seed.

[Rory Breaker]

I'm sorry but I can't let this one be. I've grouped every gun since 1.4 and I can tell u that these "targeting anilasise" are bull. The best way to group any gun is set a lan in dust or any other map with some of those good ol crates in them and play alone (0 ping, no teams), and do your bursts at a chosen specific range at the X in the crate. You'll quickly learn the accuracy of each weapon in any situation. I guess I've got myself to blame for not typing out my earlier findings earlier (or getting the support for such a page...).