[yokomo]

Originally Posted by AbraCadabra I have exact same problem. Can anybody help us? build 6027 metamod 1.21p37 amxmodx 1.8.3

That's why i don't use 6027, the recoil and bullet reg is a bit weird. Build 5758 works fine, but i can't reproduce the bug like in demo posted by Numb.

[fcastro16]

how do I downgrade?

[ANTICHRISTUS]

Originally Posted by fcastro16 ... When I use the lastest engine from hldsupdatetool, the registry is fine. Does anybody have an idea?

report that bug to Valve.

[MPNumB]

The accuracy calculation glitch I have already reported to valve, and their response was, that they won't fix it. This bug still exists in the game, and this plugin still fixes it.

It's a fix for mathematical trajectory of the bullet. It has nothing to do with actual server<->client registration. But plugin does improve accuracy of the bullet by fixing a mathematical glitch in the game.

[Debesėlis]

Does this fix on new update?

PHP Code:

Protocol version 48
Exe version 1.1.2.7/Stdio (cstrike)
Exe build: 13:14:12 Aug 29 2013 (6153) 


Doesn't fix on my server.

[MPNumB]

Still works when I test it. Have you looked at the demos, and try to replicate it with plugin on and off, while with cl_lw 0? Please do that. If after ak47 spray (like 25 bullets in a row) with plugin enabled and cl_lw set to 0, your first shot after the spray (don't reload, don't switch weapons) will be off the target, than you probably done something wrong.

[Debesėlis]

What does this command do?

PHP Code:

cl_lw 0 


I test plugin on cl_lw 1.

[MPNumB]

cl_lw 0 forces server-side weapon animation and bullet decals. It disables client weapon predictions. However this command isn't recommended, due to ping, cause then weapons may seem as if acting with some delay. It's useful only for testing of what is actually going on with the server.

[Hanzoligen]

Edit, 28.01.2014: While writing this section I did not know enough of the game mechanics and got a bit confused. I cleared the single shot behaviour in my post below. I leave the original text as a reference. Please see my corrections below.
As an old school CS 1.6 player I think this is highly interesting because I thought by now every aspect of this game is figured out.

Just to get this right. So you say that without reloading or weapon switching single shots (!) will be calculated 1, 2, 3, 4, 5, 6, 7 instead of 1, 1, 1, 1, 1, 1, 1.
Of course there is no pitch applied when using single shot, but this leads to the shots not being in the very center after you shot your first shot, is that correct?
I know the recoil pattern is not completely random because pro-gaming people have figured out what the recoil pattern look like and try to counter them to get as near to pin-point accuracy as they can by moving the mouse in the opposed direction.
If I got your explanation correct then after spraying 25 shots with the AK (for example) you - in fact - place your 26th shot after that false cooldown and this does not have the same accuracy as your first shot. When switching weapons or reloading your do have a correct cooldown and start to begin with shot number one again.

Of course because of the improved HL netcode (introduced in HL v1.1.1.0 back in 2001 along with CS Beta 6.5 and still present to this day) has lag compensation and fire animations and recoil are calculated on the client-side by default (cl_lw = 1) and only assume where the weapon hits while the server calculates the exact hits and sends them back to the client. With cl_lw = 0 you can see the actual hits coming from the server rather than the predicted hits calculated on your client. Setting cl_lw = 0 should lead to a noticeable input lag upon firing the weapon depending on your latency.

Is my explanation correct? Because if it is, I still have one question left... How come people playing on "aim_"-maps didn't complain about single shots being inaccurate? A lot of gamers try to place single shots with a wepon like the AK and when I got you right these should be anything but pin-point accurate (more like a good old spray without the pitch).

Edit, 28.01.2014: As I was digging deeper into this topic I discovered this post from a valve official:
http://forums.steampowered.com/forum....php?t=1572933
In bullet point one he is referring to exactly this behaviour of (undesired) weapon inaccuracy after you gained accuracy penalty after making use of burst fire (aka "spray and prey") or quickly repeated single shots (they also seem to count as burst in this case). Albeit the post being in the CSS section of the forum he is mentioning that the weapon recoil behaviour exists "most likely" since CS 1.6 if I got this fact correctly. Therefore I conclude that this behaviour of weapon recoil does NOT occur when you single tap the mouse and fire one shot after the other (at least not when you don't exceed a given frequency until which the shots are simply counted as single shots and not "penalty-worthy"). This explains my question above and corrects my assumptions as this was just not accurate enough how the game mechanic works. To sum it up I believe this bug only occurs when the spread penalty is high enough and you don't reload or switch weapons (of course). This would also explain why this was undetected most of the time since a lot of players either single tap the mouse or tend to reload/switch to knife/secondary quite fast during combat.


...man I would LOVE to see the source code for Counter-Strike 1.6 weapon recoil. I so much like to see how the formula looks and what amount of randomness is added. Any hints are highly appreciated.

[MPNumB]

I'll try to answer as I read through.

If you just switched or reloaded your weapon, and you are firing single shots, and waiting for recoil to cool down, it's like this:

1 (should be 1), 2 (should be 1), 2 (should be 1), 2 (should be 1), 2 (should be 1), 2 (should be 1)...

It will get back to 1 only if you reload or switch weapons. If you are firing non-stop it's after reload or switch:

1 (1), 3 (2), 4 (3), 5 (4) ...wait for cool-down... 6 (should be 1), 3 (should be 2), 4 (3)...

Also it's a bit more complicated than that, cause there still is 1 number missing, but I don't want to go into that now - will be even more complicated, cause actually engine goes from 1 to 4... But I'll fix that in next update, and will just make it as "FIXED spread transition is smoother", cause it will be really complicated to explain.

There is no patch - only fix is to switch weapons twice and/or reload (depends on the weapon), or for server to have this plugin installed.

Recoil itself is just twice the punch you see on your screen. However it does change left/right directions based on random number generator, which chance is different for each weapon in each condition (on ground, moving, ducked, ect). And then there is the spread of bullet - how much off it is from the recoil center (you can see it in my demo example when using sg550, the recoil is cooled down between shots, however the spread is still not dead-center, cause it isn't cooled-down). Recoil is easy to predict - twice the punchangle, and adapt to direction change. Spread however isn't as easy.

Yes, normally switching weapons and/or reloading (based on the weapon), engine refreshes your shot accuracy. And my plugin fixes that automatically, no need to reload or switch weapons.

Yes cl_lw 1 is only client trying to predict where the bullet will hit, but the real hit location where the damage is taking place - that's cl_lw 0. So yes, what you said about cl_lw is accurate. However I have noticed that when you set it to 0, it still sometimes has package loss. I tested it out with n0br41ner, and we noticed that sometimes either the package is lost, or client is attempting to predict the bullet location anyways. For example in one of the demos I've uploaded (they were both recorded with cl_lw 0), in sg550 when watching first time you see one spray patter, and when watching an other time, you'll see slightly different one (one or two bullet locations were sent from the server to the client during the recording).

And the reason for why people aren't complaining much about it in aim maps when single-shooting, my guess it's cause difference in spread between 1st and 3rd bullet isn't really big. However it still is enough to make a difference at a distance.

From what I understood... Well, the glitch happens even if you're making single shots. The more spread you had on your last bullet, the bigger the effect. However... this glitch was well known (in my opinion) (at least when I was playing CS1.5), and it is the source of why some people started to switch weapons twice after shooting with ak47 or other large-spread-weapons. Other people just copied this strategy without even knowing why, and that's how it passed to be now. I'm basing this, cause I had friend in the old days who liked ak47, and when asked why he's switching weapons so often, he just told "cause else ak isn't accurate".

And about the code of weapons, you may want to take a look here:
https://github.com/Arkshine/CSSDK/tr...lls/wpn_shared

Also will update the plugin soon. I already made the code, but right now I'm in a rush to do some things. Here's what to expect:

 *  + 2.0
 *  - Added: Fix for smooth spread transitions of automatic weapons.
 *  - Added: Improved smooth spread transitions of automatic weapons.
 *  - Added: Fix for slow spread cool-down of automatic weapons at low fps.
 *  - Added: Smoother transition to maximum accuracy for pistols and sg550.