Articulo sacado del foro del steam
http://forums.steampowered.com/forums/showthread.php?threadid=490324
Content:
1) Introduction
2) Test environment
3) The tests
3.1) cl_interpolate 0/1
3.2) cl_interpolate & cl_lagcompensation
3.3) cl_interpolate & cl_interp
4) Conclusion
1) Introduction:
Hello. My name is Christoph ‚disruptor’ Dölger. I’m a German professional cyber-athlete and I’m playing for the international known Clan ‚n!faculty’ (www.faculty.de). I’m playing Counter-Strike since November 1999 (Beta 4) and Counter-Strike:Source since February 2005. I’m just waiting for a net-code-fix by valve, too long. I posted this thread, because I want Valve to recognize, that many Gamers are dissatisfied with the current situation. This thread is not mentioned to show how the net-code work or how the net-code should work. The point is that Valve has to recognize the problem with the net-code and to fix it! This thread is for all the cyber-athletes that see CS:S as a sport and matching each other on tournaments and leagues.
The game is because of the current net-code-bugs NOT competitive!
2) Test environment:
I accomplished my tests on two “high-end”-Computers and one dedicated LAN-Server. The Computers reached on every single test 100fps. The dedicated LAN-Server was located in my personal home-network and provided a constant 100-tickrate. The general net-settings of the clients were: rate 25000, cl_cmdrate 101 and cl_updaterate 101. These Values are forced on LANs and events. I accomplished my tests on LAN because of the low-latency and the quick Client-Server-Communication. However the results of the tests are transferable to the internet and (nearly) identical! But on LAN you still have the best premises.
3) The tests:
3.1) cl_interpolate 0/1:
First of all you should watch the following video several times:
http://momo.coxcomb-worx.de/disi/hi...erpolate1-0.wmv
On that video you can recognize, that at “cl_interpolate 1” the model that stands still, sees the moving model delayed. In this situation, as a waiting/camping player you have a reaction-disadvantage, because the moving model sees the standing model sooner and because of that the moving model can react faster than the camping player!
At “cl_interpolate 0” you are seeing the moving model at the same time as the moving model sees you! So there is neither a reaction advantage nor a reaction disadvantage at “cl_interpolate 0”!!
The next image makes that clear:
http://momo.coxcomb-worx.de/disi/hi...erpolate1-0.jpg
Result:
You should assume, that “cl_interpolate 0” is the best setting, BUT there is a bug in the lagcompensation (see below).
3.2) cl_interpolate & cl_lagcompensation:
cl_interpolate 0 & cl_lagcompensation 1:
These settings are bad, because you have to shoot behind the model to hit it.
http://momo.coxcomb-worx.de/disi/hitbox_vid_ip0_lc1.wmv
http://momo.coxcomb-worx.de/disi/hitbox_pic_ip0_lc1.jpg
cl_interpolate 0 & cl_lagcompensation 0:
These settings are perfect! The hit-boxes are on the model and because of “cl_interpolate 0” nobody has a reaction-disadvantage! But you can ONLY use these settings on LAN (because of the low-latency), because the lag-compensation is disabled!
http://momo.coxcomb-worx.de/disi/hitbox_vid_ip0_lc0.wmv
http://momo.coxcomb-worx.de/disi/hitbox_pic_ip0_lc0.jpg
cl_interpolate 1 & cl_lagcompensation 1:
These settings are forced on LAN tournaments (unfortunately)… and on internet leagues. But they are also bad, because of the reaction-disadvantage by “cl_interpolate 1”!!!
http://momo.coxcomb-worx.de/disi/hitbox_vid_ip1_lc1.wmv
http://momo.coxcomb-worx.de/disi/hitbox_pic_ip1_lc1.jpg
cl_interpolate 1 & cl_lagcompensation 0:
These settings are just too bad, because you have to shoot in front of the model to hit it!
http://momo.coxcomb-worx.de/disi/hitbox_vid_ip1_lc0.wmv
http://momo.coxcomb-worx.de/disi/hitbox_pic_ip1_lc0.jpg
Result:
As you can see above, the best client-settings are clearly “cl_interpolate 0” AND “cl_lagcompensation 0”, BUT you can ONLY use these settings on LAN, because the lag-compensation is (client-sided) disabled! However this is no problem on LAN, because every player has got a ping under 15ms!
On the internet the players has got a ping from 25ms and higher. On that “high-ping”-gamers and the settings “cl_interpolate 0” AND “cl_lagcompensation 0”, you have to shoot in FRONT of the model so that it is not adaptable to the internet! And because of that, we all have to play on “cl_interpolate 1” AND “cl_lagcompensation 1” online! That’s just the lesser of two evils!
3.3) cl_interpolate & cl_interp:
If you take a look at the picture:
http://momo.coxcomb-worx.de/disi/hitbox_pic_interp.jpg
You should recognize that the “cl_interp”-Value doesn’t matter on “cl_interpolate 0”, because the lag-compensation is disabled!
The higher the “cl_interp”-Value is chosen at “cl_interpolate 1” the later you see the model!
Result:
The setting “cl_interp 0.01” is obviously the best.
4) Conclusion:
My conclusion is easy:
On LAN with a tickrate-100-server you have to configure the following client-settings: cl_interpolate 0, cl_lagcompensation 0, cl_interp 0.01 and the following server-setting: sv_unlag 0 (“0” = disables player lag-compensation).
Unfortunately, we all have to play with the lesser of two evils in the internet (that means: “cl_interpolate 1”, “cl_lagcompensation 1”)! And here Valve comes into play!
VALVE! PLEASE FINALLY FIX THE LAG-COMPENSATION!!!!!!!!!
P.D. Si alguien quire traducirlo 
resumiendo en LAN interpolate 0 Inet interpolate 1
